diff --git "a/wandb/run-20220405_210343-1w1tow7f/files/wandb-summary.json" "b/wandb/run-20220405_210343-1w1tow7f/files/wandb-summary.json"
--- "a/wandb/run-20220405_210343-1w1tow7f/files/wandb-summary.json"
+++ "b/wandb/run-20220405_210343-1w1tow7f/files/wandb-summary.json"
@@ -1 +1 @@
-{"train/decoder_grad_norm": 6.248335361480713, "train/decoder_param_norm": 1043.5452880859375, "train/encoder_grad_norm": 3.8687353134155273, "train/encoder_param_norm": 2309.398681640625, "train/grad_norm": 7.349069595336914, "layer_grad_norm/": {"decoder": {"model": {"decoder": {"embed_positions": {"embedding": 0.3957214951515198}, "embed_tokens": {"embedding": 0.957571804523468}, "layernorm_embedding": {"bias": 0.19529378414154053, "scale": 0.1615617871284485}, "layers": {"0": {"encoder_attn": {"k_proj": {"bias": 6.126289463281864e-06, "kernel": 0.00512276915833354}, "out_proj": {"bias": 0.06256742775440216, "kernel": 0.11431083083152771}, "q_proj": {"bias": 0.00020890458836220205, "kernel": 0.00432879664003849}, "v_proj": {"bias": 0.1571826934814453, "kernel": 0.07774726301431656}}, "encoder_attn_layer_norm": {"bias": 0.09683283418416977, "scale": 0.08474510163068771}, "fc1": {"bias": 0.04747888445854187, "kernel": 0.7181474566459656}, "fc2": {"bias": 0.07666343450546265, "kernel": 0.7600010633468628}, "final_layer_norm": {"bias": 0.23582476377487183, "scale": 0.10399508476257324}, "self_attn": {"k_proj": {"bias": 2.0379340639919974e-05, "kernel": 0.11543459445238113}, "out_proj": {"bias": 0.10107098519802094, "kernel": 0.7000217437744141}, "q_proj": {"bias": 0.013378174044191837, "kernel": 0.16752809286117554}, "v_proj": {"bias": 0.11671453714370728, "kernel": 0.7328258156776428}}, "self_attn_layer_norm": {"bias": 0.059811219573020935, "scale": 0.05270637944340706}}, "1": {"encoder_attn": {"k_proj": {"bias": 3.88754187952145e-06, "kernel": 0.0030759149231016636}, "out_proj": {"bias": 0.06340594589710236, "kernel": 0.11241766065359116}, "q_proj": {"bias": 0.0001222345163114369, "kernel": 0.0030684275552630424}, "v_proj": {"bias": 0.14297497272491455, "kernel": 0.07018139213323593}}, "encoder_attn_layer_norm": {"bias": 0.10328488051891327, "scale": 0.08806563913822174}, "fc1": {"bias": 0.04447054862976074, "kernel": 0.6831641793251038}, "fc2": {"bias": 0.08077238500118256, "kernel": 0.7472724318504333}, "final_layer_norm": {"bias": 0.26292380690574646, "scale": 0.15282970666885376}, "self_attn": {"k_proj": {"bias": 1.6001253243302926e-05, "kernel": 0.10096275806427002}, "out_proj": {"bias": 0.12635289132595062, "kernel": 0.37729811668395996}, "q_proj": {"bias": 0.011176313273608685, "kernel": 0.1370301991701126}, "v_proj": {"bias": 0.14967839419841766, "kernel": 0.4089428782463074}}, "self_attn_layer_norm": {"bias": 0.06022721529006958, "scale": 0.05527588352560997}}, "10": {"encoder_attn": {"k_proj": {"bias": 3.2692237255105283e-06, "kernel": 0.00407384755089879}, "out_proj": {"bias": 0.07940680533647537, "kernel": 0.11943644285202026}, "q_proj": {"bias": 0.00024411304912064224, "kernel": 0.005123197101056576}, "v_proj": {"bias": 0.15195327997207642, "kernel": 0.07468514144420624}}, "encoder_attn_layer_norm": {"bias": 0.12791143357753754, "scale": 0.12387944757938385}, "fc1": {"bias": 0.051502399146556854, "kernel": 2.3145861625671387}, "fc2": {"bias": 0.1201619878411293, "kernel": 1.014142394065857}, "final_layer_norm": {"bias": 1.1386778354644775, "scale": 0.19159001111984253}, "self_attn": {"k_proj": {"bias": 1.3440743714454584e-05, "kernel": 0.10177838057279587}, "out_proj": {"bias": 0.34463241696357727, "kernel": 0.28363943099975586}, "q_proj": {"bias": 0.008246778510510921, "kernel": 0.11239436268806458}, "v_proj": {"bias": 0.36987680196762085, "kernel": 0.9525725841522217}}, "self_attn_layer_norm": {"bias": 0.07684528082609177, "scale": 0.06492282450199127}}, "11": {"encoder_attn": {"k_proj": {"bias": 6.272938662732486e-06, "kernel": 0.005256214644759893}, "out_proj": {"bias": 0.09759582579135895, "kernel": 0.12459102272987366}, "q_proj": {"bias": 0.0002283074427396059, "kernel": 0.005217547528445721}, "v_proj": {"bias": 0.19665828347206116, "kernel": 0.09632201492786407}}, "encoder_attn_layer_norm": {"bias": 0.12574829161167145, "scale": 0.09106750786304474}, "fc1": {"bias": 0.07080992311239243, "kernel": 2.785266637802124}, "fc2": {"bias": 0.1415998488664627, "kernel": 1.281799554824829}, "final_layer_norm": {"bias": 0.11112289130687714, "scale": 0.15087562799453735}, "self_attn": {"k_proj": {"bias": 1.0508645573281683e-05, "kernel": 0.07647208869457245}, "out_proj": {"bias": 0.743054211139679, "kernel": 0.2663988471031189}, "q_proj": {"bias": 0.006267210468649864, "kernel": 0.08555609732866287}, "v_proj": {"bias": 0.7096261978149414, "kernel": 0.552697479724884}}, "self_attn_layer_norm": {"bias": 0.0937671884894371, "scale": 0.08987914770841599}}, "2": {"encoder_attn": {"k_proj": {"bias": 2.9370780794124585e-06, "kernel": 0.003957669250667095}, "out_proj": {"bias": 0.07270092517137527, "kernel": 0.12159011512994766}, "q_proj": {"bias": 0.0001965147239388898, "kernel": 0.0034371146466583014}, "v_proj": {"bias": 0.1693577766418457, "kernel": 0.08288073539733887}}, "encoder_attn_layer_norm": {"bias": 0.10082058608531952, "scale": 0.09459296613931656}, "fc1": {"bias": 0.043445006012916565, "kernel": 0.6674180626869202}, "fc2": {"bias": 0.08264637738466263, "kernel": 0.7661255598068237}, "final_layer_norm": {"bias": 0.2397005558013916, "scale": 0.17274972796440125}, "self_attn": {"k_proj": {"bias": 1.6519239579793066e-05, "kernel": 0.10892654210329056}, "out_proj": {"bias": 0.13029983639717102, "kernel": 0.37324243783950806}, "q_proj": {"bias": 0.01230115070939064, "kernel": 0.13011199235916138}, "v_proj": {"bias": 0.16930969059467316, "kernel": 0.4702507555484772}}, "self_attn_layer_norm": {"bias": 0.0712442398071289, "scale": 0.0684252679347992}}, "3": {"encoder_attn": {"k_proj": {"bias": 2.6690172489907127e-06, "kernel": 0.002852358855307102}, "out_proj": {"bias": 0.06669719517230988, "kernel": 0.0965690016746521}, "q_proj": {"bias": 0.00012008623889414594, "kernel": 0.0027435708325356245}, "v_proj": {"bias": 0.130833700299263, "kernel": 0.06440014392137527}}, "encoder_attn_layer_norm": {"bias": 0.10752736777067184, "scale": 0.09413988888263702}, "fc1": {"bias": 0.03976931422948837, "kernel": 0.6132485270500183}, "fc2": {"bias": 0.08255304396152496, "kernel": 0.7024412155151367}, "final_layer_norm": {"bias": 0.28141605854034424, "scale": 0.11679420620203018}, "self_attn": {"k_proj": {"bias": 2.1457493858179078e-05, "kernel": 0.13284903764724731}, "out_proj": {"bias": 0.12292786687612534, "kernel": 0.2719310522079468}, "q_proj": {"bias": 0.01649159938097, "kernel": 0.15851031243801117}, "v_proj": {"bias": 0.15283986926078796, "kernel": 0.3871537446975708}}, "self_attn_layer_norm": {"bias": 0.06379645317792892, "scale": 0.06390906870365143}}, "4": {"encoder_attn": {"k_proj": {"bias": 3.3133851502498146e-06, "kernel": 0.0035838205367326736}, "out_proj": {"bias": 0.07109887897968292, "kernel": 0.09054248034954071}, "q_proj": {"bias": 0.00015105115016922355, "kernel": 0.003207859117537737}, "v_proj": {"bias": 0.1483682543039322, "kernel": 0.07266827672719955}}, "encoder_attn_layer_norm": {"bias": 0.10147704929113388, "scale": 0.08346009254455566}, "fc1": {"bias": 0.040273673832416534, "kernel": 0.6174589991569519}, "fc2": {"bias": 0.08745791018009186, "kernel": 0.7280296087265015}, "final_layer_norm": {"bias": 0.32406747341156006, "scale": 0.15126483142375946}, "self_attn": {"k_proj": {"bias": 1.5435152818099596e-05, "kernel": 0.12396755069494247}, "out_proj": {"bias": 0.13306371867656708, "kernel": 0.2527622580528259}, "q_proj": {"bias": 0.012039960362017155, "kernel": 0.15329518914222717}, "v_proj": {"bias": 0.1754838526248932, "kernel": 0.33414921164512634}}, "self_attn_layer_norm": {"bias": 0.06894379109144211, "scale": 0.06161382049322128}}, "5": {"encoder_attn": {"k_proj": {"bias": 2.579424290161114e-06, "kernel": 0.003295755013823509}, "out_proj": {"bias": 0.07313599437475204, "kernel": 0.08332156389951706}, "q_proj": {"bias": 0.00016511954891029745, "kernel": 0.0031555830501019955}, "v_proj": {"bias": 0.1671561747789383, "kernel": 0.08049289137125015}}, "encoder_attn_layer_norm": {"bias": 0.10280616581439972, "scale": 0.08457314968109131}, "fc1": {"bias": 0.03733683004975319, "kernel": 0.5850957036018372}, "fc2": {"bias": 0.09051179140806198, "kernel": 0.7403404116630554}, "final_layer_norm": {"bias": 0.2841578423976898, "scale": 0.1211640015244484}, "self_attn": {"k_proj": {"bias": 2.663955274329055e-05, "kernel": 0.13452614843845367}, "out_proj": {"bias": 0.1335134506225586, "kernel": 0.26720350980758667}, "q_proj": {"bias": 0.010996587574481964, "kernel": 0.15277327597141266}, "v_proj": {"bias": 0.1855141669511795, "kernel": 0.3694325387477875}}, "self_attn_layer_norm": {"bias": 0.06823275238275528, "scale": 0.05866802856326103}}, "6": {"encoder_attn": {"k_proj": {"bias": 3.3172398161696037e-06, "kernel": 0.003980656154453754}, "out_proj": {"bias": 0.07266882807016373, "kernel": 0.0909934937953949}, "q_proj": {"bias": 0.0001784397172741592, "kernel": 0.003545940387994051}, "v_proj": {"bias": 0.14963947236537933, "kernel": 0.07267127931118011}}, "encoder_attn_layer_norm": {"bias": 0.10383890569210052, "scale": 0.08422869443893433}, "fc1": {"bias": 0.03749598562717438, "kernel": 0.6408689618110657}, "fc2": {"bias": 0.08706986159086227, "kernel": 0.7343899011611938}, "final_layer_norm": {"bias": 0.21194224059581757, "scale": 0.1415112167596817}, "self_attn": {"k_proj": {"bias": 2.4872339054127224e-05, "kernel": 0.14111733436584473}, "out_proj": {"bias": 0.13314063847064972, "kernel": 0.2901186943054199}, "q_proj": {"bias": 0.013028834946453571, "kernel": 0.16716249287128448}, "v_proj": {"bias": 0.17111794650554657, "kernel": 0.4056183695793152}}, "self_attn_layer_norm": {"bias": 0.07046671956777573, "scale": 0.06165720522403717}}, "7": {"encoder_attn": {"k_proj": {"bias": 2.9195082333899336e-06, "kernel": 0.003328036516904831}, "out_proj": {"bias": 0.06549763679504395, "kernel": 0.07904414087533951}, "q_proj": {"bias": 0.00016097885963972658, "kernel": 0.003293772228062153}, "v_proj": {"bias": 0.13465145230293274, "kernel": 0.06527221202850342}}, "encoder_attn_layer_norm": {"bias": 0.10720624774694443, "scale": 0.08830474317073822}, "fc1": {"bias": 0.03968070447444916, "kernel": 0.6990951895713806}, "fc2": {"bias": 0.09039907157421112, "kernel": 0.7943719625473022}, "final_layer_norm": {"bias": 0.20924066007137299, "scale": 0.1474452167749405}, "self_attn": {"k_proj": {"bias": 1.9321842046338134e-05, "kernel": 0.11917725950479507}, "out_proj": {"bias": 0.12267930060625076, "kernel": 0.3357153534889221}, "q_proj": {"bias": 0.010344602167606354, "kernel": 0.1380189210176468}, "v_proj": {"bias": 0.12153981626033783, "kernel": 0.4702933132648468}}, "self_attn_layer_norm": {"bias": 0.06426583975553513, "scale": 0.0607207827270031}}, "8": {"encoder_attn": {"k_proj": {"bias": 2.6780580810736865e-06, "kernel": 0.002897089347243309}, "out_proj": {"bias": 0.0653957650065422, "kernel": 0.09783539175987244}, "q_proj": {"bias": 0.00014650513185188174, "kernel": 0.0030433807987719774}, "v_proj": {"bias": 0.12528745830059052, "kernel": 0.06073777750134468}}, "encoder_attn_layer_norm": {"bias": 0.10408639907836914, "scale": 0.10603990405797958}, "fc1": {"bias": 0.044246166944503784, "kernel": 0.6812352538108826}, "fc2": {"bias": 0.0909360945224762, "kernel": 0.8466796875}, "final_layer_norm": {"bias": 0.21654388308525085, "scale": 0.11097487062215805}, "self_attn": {"k_proj": {"bias": 1.384177721774904e-05, "kernel": 0.0956977903842926}, "out_proj": {"bias": 0.11662668734788895, "kernel": 0.31024736166000366}, "q_proj": {"bias": 0.007025822065770626, "kernel": 0.10700628906488419}, "v_proj": {"bias": 0.12875394523143768, "kernel": 0.5613791942596436}}, "self_attn_layer_norm": {"bias": 0.06248074397444725, "scale": 0.0605536587536335}}, "9": {"encoder_attn": {"k_proj": {"bias": 2.759453991529881e-06, "kernel": 0.0030704366508871317}, "out_proj": {"bias": 0.06583196669816971, "kernel": 0.11220739036798477}, "q_proj": {"bias": 0.0001448578404961154, "kernel": 0.003283187048509717}, "v_proj": {"bias": 0.12293069809675217, "kernel": 0.06001891568303108}}, "encoder_attn_layer_norm": {"bias": 0.10925756394863129, "scale": 0.1169716939330101}, "fc1": {"bias": 0.04894828423857689, "kernel": 1.167492151260376}, "fc2": {"bias": 0.0926441177725792, "kernel": 0.9564423561096191}, "final_layer_norm": {"bias": 0.5103340744972229, "scale": 0.19271788001060486}, "self_attn": {"k_proj": {"bias": 1.2695875739154872e-05, "kernel": 0.09514088928699493}, "out_proj": {"bias": 0.14068712294101715, "kernel": 0.2597719132900238}, "q_proj": {"bias": 0.006812964100390673, "kernel": 0.10133656859397888}, "v_proj": {"bias": 0.15058520436286926, "kernel": 0.7108777165412903}}, "self_attn_layer_norm": {"bias": 0.06411344558000565, "scale": 0.05926334112882614}}}}}}, "encoder": {"adapter": {"layers": {"0": {"conv": {"bias": 0.20907211303710938, "kernel": 1.3787331581115723}}, "1": {"conv": {"bias": 0.3797065317630768, "kernel": 1.2682850360870361}}, "2": {"conv": {"bias": 0.7108349204063416, "kernel": 1.1133043766021729}}}}, "encoder": {"layer_norm": {"bias": 0.22636950016021729, "scale": 0.16822795569896698}, "layers": {"0": {"attention": {"k_proj": {"bias": 6.208925333339721e-05, "kernel": 0.17955805361270905}, "out_proj": {"bias": 0.07983482629060745, "kernel": 0.7962406277656555}, "q_proj": {"bias": 0.06279527395963669, "kernel": 0.20784415304660797}, "v_proj": {"bias": 0.36112138628959656, "kernel": 0.6930927634239197}}, "feed_forward": {"intermediate_dense": {"bias": 0.11735843122005463, "kernel": 0.7532103657722473}, "output_dense": {"bias": 0.02027445286512375, "kernel": 0.8071455955505371}}, "final_layer_norm": {"bias": 0.6862912774085999, "scale": 0.7094218134880066}, "layer_norm": {"bias": 0.9261000752449036, "scale": 0.9945785403251648}}, "1": {"attention": {"k_proj": {"bias": 4.656344390241429e-05, "kernel": 0.1650291234254837}, "out_proj": {"bias": 0.02179878205060959, "kernel": 0.3399544954299927}, "q_proj": {"bias": 0.025889579206705093, "kernel": 0.15403114259243011}, "v_proj": {"bias": 0.08082185685634613, "kernel": 0.2983188331127167}}, "feed_forward": {"intermediate_dense": {"bias": 0.07131170481443405, "kernel": 0.5860697627067566}, "output_dense": {"bias": 0.01132867019623518, "kernel": 0.2675647735595703}}, "final_layer_norm": {"bias": 0.4152465760707855, "scale": 0.42845815420150757}, "layer_norm": {"bias": 0.3494342267513275, "scale": 0.3329411745071411}}, "10": {"attention": {"k_proj": {"bias": 1.3912446092945174e-06, "kernel": 0.005678324960172176}, "out_proj": {"bias": 0.0012141831684857607, "kernel": 0.01808365061879158}, "q_proj": {"bias": 0.0014122046995908022, "kernel": 0.005739105399698019}, "v_proj": {"bias": 0.005440743640065193, "kernel": 0.017127597704529762}}, "feed_forward": {"intermediate_dense": {"bias": 0.0005559786222875118, "kernel": 0.015033891424536705}, "output_dense": {"bias": 0.0011165253818035126, "kernel": 0.013163355179131031}}, "final_layer_norm": {"bias": 0.0026556658558547497, "scale": 0.00272438395768404}, "layer_norm": {"bias": 0.02731502428650856, "scale": 0.023141326382756233}}, "11": {"attention": {"k_proj": {"bias": 1.4308424169939826e-06, "kernel": 0.005131026729941368}, "out_proj": {"bias": 0.0010771803790703416, "kernel": 0.01833004504442215}, "q_proj": {"bias": 0.0009028510539792478, "kernel": 0.0053285458125174046}, "v_proj": {"bias": 0.00486570643261075, "kernel": 0.016590269282460213}}, "feed_forward": {"intermediate_dense": {"bias": 0.0004334738478064537, "kernel": 0.015502448193728924}, "output_dense": {"bias": 0.0010182925034314394, "kernel": 0.012219805270433426}}, "final_layer_norm": {"bias": 0.0017391270957887173, "scale": 0.0022147661074995995}, "layer_norm": {"bias": 0.023472534492611885, "scale": 0.015590372495353222}}, "12": {"attention": {"k_proj": {"bias": 1.3596761618828168e-06, "kernel": 0.005960967391729355}, "out_proj": {"bias": 0.0008078793762251735, "kernel": 0.017825184389948845}, "q_proj": {"bias": 0.0009535121498629451, "kernel": 0.006382802966982126}, "v_proj": {"bias": 0.003964895848184824, "kernel": 0.016368791460990906}}, "feed_forward": {"intermediate_dense": {"bias": 0.0013143799733370543, "kernel": 0.019410952925682068}, "output_dense": {"bias": 0.0007427112432196736, "kernel": 0.01928628608584404}}, "final_layer_norm": {"bias": 0.005945633165538311, "scale": 0.009220371022820473}, "layer_norm": {"bias": 0.019585484638810158, "scale": 0.01862788200378418}}, "13": {"attention": {"k_proj": {"bias": 1.1747559938157792e-06, "kernel": 0.005064372438937426}, "out_proj": {"bias": 0.000696739531122148, "kernel": 0.015778791159391403}, "q_proj": {"bias": 0.0008420691010542214, "kernel": 0.005768876057118177}, "v_proj": {"bias": 0.002978556789457798, "kernel": 0.012962739914655685}}, "feed_forward": {"intermediate_dense": {"bias": 0.0009936989517882466, "kernel": 0.018645890057086945}, "output_dense": {"bias": 0.0006305691204033792, "kernel": 0.01773606799542904}}, "final_layer_norm": {"bias": 0.004375901073217392, "scale": 0.0062483069486916065}, "layer_norm": {"bias": 0.01600462943315506, "scale": 0.010493500158190727}}, "14": {"attention": {"k_proj": {"bias": 1.0404260137875099e-06, "kernel": 0.004872146528214216}, "out_proj": {"bias": 0.0006074492121115327, "kernel": 0.01399173028767109}, "q_proj": {"bias": 0.0006728185107931495, "kernel": 0.004937955178320408}, "v_proj": {"bias": 0.0024946907069534063, "kernel": 0.011828620918095112}}, "feed_forward": {"intermediate_dense": {"bias": 0.0010532960295677185, "kernel": 0.021244361996650696}, "output_dense": {"bias": 0.0005741144996136427, "kernel": 0.01359262689948082}}, "final_layer_norm": {"bias": 0.004819571040570736, "scale": 0.0058192419819533825}, "layer_norm": {"bias": 0.012392263859510422, "scale": 0.00963513646274805}}, "15": {"attention": {"k_proj": {"bias": 1.2368124089334742e-06, "kernel": 0.005688127130270004}, "out_proj": {"bias": 0.0005401051021181047, "kernel": 0.014547410421073437}, "q_proj": {"bias": 0.0008324682130478323, "kernel": 0.006829832214862108}, "v_proj": {"bias": 0.0023668180219829082, "kernel": 0.01182851754128933}}, "feed_forward": {"intermediate_dense": {"bias": 0.0010811567772179842, "kernel": 0.016164839267730713}, "output_dense": {"bias": 0.0005189789808355272, "kernel": 0.014824673533439636}}, "final_layer_norm": {"bias": 0.00506795896217227, "scale": 0.005178646184504032}, "layer_norm": {"bias": 0.011762675829231739, "scale": 0.01217866875231266}}, "16": {"attention": {"k_proj": {"bias": 1.061342459252046e-06, "kernel": 0.005670086480677128}, "out_proj": {"bias": 0.0004969135043211281, "kernel": 0.015095161274075508}, "q_proj": {"bias": 0.0006495080306194723, "kernel": 0.00609693955630064}, "v_proj": {"bias": 0.0020813094452023506, "kernel": 0.011373939923942089}}, "feed_forward": {"intermediate_dense": {"bias": 0.0009626346291042864, "kernel": 0.017702966928482056}, "output_dense": {"bias": 0.00047558321966789663, "kernel": 0.01638943701982498}}, "final_layer_norm": {"bias": 0.004663620609790087, "scale": 0.0040918393060564995}, "layer_norm": {"bias": 0.011288191191852093, "scale": 0.00861026719212532}}, "17": {"attention": {"k_proj": {"bias": 1.1684753644658485e-06, "kernel": 0.005791741888970137}, "out_proj": {"bias": 0.0004660442937165499, "kernel": 0.01514619030058384}, "q_proj": {"bias": 0.0007063922239467502, "kernel": 0.00639334274455905}, "v_proj": {"bias": 0.0018986751092597842, "kernel": 0.01108721923083067}}, "feed_forward": {"intermediate_dense": {"bias": 0.0012936763232573867, "kernel": 0.026417791843414307}, "output_dense": {"bias": 0.0003913144173566252, "kernel": 0.019519170746207237}}, "final_layer_norm": {"bias": 0.006019943859428167, "scale": 0.00761314295232296}, "layer_norm": {"bias": 0.0109083391726017, "scale": 0.008430450223386288}}, "18": {"attention": {"k_proj": {"bias": 1.1292753470115713e-06, "kernel": 0.008385256864130497}, "out_proj": {"bias": 0.0003829613560810685, "kernel": 0.01349470391869545}, "q_proj": {"bias": 0.0009535849676467478, "kernel": 0.00888859760016203}, "v_proj": {"bias": 0.0018109718803316355, "kernel": 0.011800557374954224}}, "feed_forward": {"intermediate_dense": {"bias": 0.0007912518340162933, "kernel": 0.021111397072672844}, "output_dense": {"bias": 0.0003573457361198962, "kernel": 0.018499381840229034}}, "final_layer_norm": {"bias": 0.0034502455964684486, "scale": 0.0033679131884127855}, "layer_norm": {"bias": 0.012115299701690674, "scale": 0.011664382182061672}}, "19": {"attention": {"k_proj": {"bias": 1.5795940271345899e-06, "kernel": 0.009211079217493534}, "out_proj": {"bias": 0.00032448535785079, "kernel": 0.013495567254722118}, "q_proj": {"bias": 0.0010837309528142214, "kernel": 0.010159092955291271}, "v_proj": {"bias": 0.0016703916480764747, "kernel": 0.011343966238200665}}, "feed_forward": {"intermediate_dense": {"bias": 0.0008422433747909963, "kernel": 0.026424439623951912}, "output_dense": {"bias": 0.0003142349305562675, "kernel": 0.021706853061914444}}, "final_layer_norm": {"bias": 0.004137103911489248, "scale": 0.004442865494638681}, "layer_norm": {"bias": 0.010162881575524807, "scale": 0.012011111713945866}}, "2": {"attention": {"k_proj": {"bias": 2.934725307568442e-05, "kernel": 0.08383528143167496}, "out_proj": {"bias": 0.012729697860777378, "kernel": 0.1819193959236145}, "q_proj": {"bias": 0.010320818983018398, "kernel": 0.07138322293758392}, "v_proj": {"bias": 0.04734956845641136, "kernel": 0.18332970142364502}}, "feed_forward": {"intermediate_dense": {"bias": 0.04062354564666748, "kernel": 0.33744746446609497}, "output_dense": {"bias": 0.008657855913043022, "kernel": 0.23138262331485748}}, "final_layer_norm": {"bias": 0.236163929104805, "scale": 0.2602633535861969}, "layer_norm": {"bias": 0.17614230513572693, "scale": 0.20887266099452972}}, "20": {"attention": {"k_proj": {"bias": 1.5735271290395758e-06, "kernel": 0.015206554904580116}, "out_proj": {"bias": 0.0002787901321426034, "kernel": 0.013272221200168133}, "q_proj": {"bias": 0.001516208634711802, "kernel": 0.015279811806976795}, "v_proj": {"bias": 0.0015550552634522319, "kernel": 0.010907074436545372}}, "feed_forward": {"intermediate_dense": {"bias": 0.0006792520871385932, "kernel": 0.025646919384598732}, "output_dense": {"bias": 0.00028622496756725013, "kernel": 0.02634827047586441}}, "final_layer_norm": {"bias": 0.0030681260395795107, "scale": 0.004423606675118208}, "layer_norm": {"bias": 0.011422278359532356, "scale": 0.010012513026595116}}, "21": {"attention": {"k_proj": {"bias": 8.310950079248869e-07, "kernel": 0.013665321283042431}, "out_proj": {"bias": 0.00020622959709726274, "kernel": 0.011521727778017521}, "q_proj": {"bias": 0.0011713471030816436, "kernel": 0.014692011289298534}, "v_proj": {"bias": 0.0012885641772300005, "kernel": 0.010312050580978394}}, "feed_forward": {"intermediate_dense": {"bias": 0.0006124781793914735, "kernel": 0.02698848769068718}, "output_dense": {"bias": 0.00023950928880367428, "kernel": 0.03338848799467087}}, "final_layer_norm": {"bias": 0.002662272658199072, "scale": 0.0030492024961858988}, "layer_norm": {"bias": 0.00962064228951931, "scale": 0.009373709559440613}}, "22": {"attention": {"k_proj": {"bias": 5.475717301806071e-08, "kernel": 0.0009106611832976341}, "out_proj": {"bias": 0.0002460938412696123, "kernel": 0.04560656473040581}, "q_proj": {"bias": 0.00017159993876703084, "kernel": 0.0018241411307826638}, "v_proj": {"bias": 0.0014203126775100827, "kernel": 0.03612879291176796}}, "feed_forward": {"intermediate_dense": {"bias": 0.0008528109174221754, "kernel": 0.057362526655197144}, "output_dense": {"bias": 0.00035851969732902944, "kernel": 0.06820782274007797}}, "final_layer_norm": {"bias": 0.0038628610782325268, "scale": 0.003719903063029051}, "layer_norm": {"bias": 0.009540154598653316, "scale": 0.013003811240196228}}, "23": {"attention": {"k_proj": {"bias": 8.878872392870107e-08, "kernel": 0.002870816271752119}, "out_proj": {"bias": 0.00036784910480491817, "kernel": 0.034839656203985214}, "q_proj": {"bias": 0.0005525403539650142, "kernel": 0.005187225993722677}, "v_proj": {"bias": 0.0031609905418008566, "kernel": 0.046591613441705704}}, "feed_forward": {"intermediate_dense": {"bias": 0.0009177798056043684, "kernel": 0.02437104471027851}, "output_dense": {"bias": 0.0003799554251600057, "kernel": 0.013147326186299324}}, "final_layer_norm": {"bias": 0.0030138962902128696, "scale": 0.0030258060432970524}, "layer_norm": {"bias": 0.03557658568024635, "scale": 0.05969923362135887}}, "3": {"attention": {"k_proj": {"bias": 1.649874320719391e-05, "kernel": 0.04841494932770729}, "out_proj": {"bias": 0.00999386701732874, "kernel": 0.11141359806060791}, "q_proj": {"bias": 0.008432073518633842, "kernel": 0.0553056076169014}, "v_proj": {"bias": 0.038721710443496704, "kernel": 0.1119893416762352}}, "feed_forward": {"intermediate_dense": {"bias": 0.029075870290398598, "kernel": 0.20446009933948517}, "output_dense": {"bias": 0.007055670954287052, "kernel": 0.12683533132076263}}, "final_layer_norm": {"bias": 0.1824246048927307, "scale": 0.19210882484912872}, "layer_norm": {"bias": 0.1613454520702362, "scale": 0.13378405570983887}}, "4": {"attention": {"k_proj": {"bias": 1.9063019863096997e-05, "kernel": 0.03732868283987045}, "out_proj": {"bias": 0.00815676525235176, "kernel": 0.0880126953125}, "q_proj": {"bias": 0.008292106911540031, "kernel": 0.043554600328207016}, "v_proj": {"bias": 0.035737793892621994, "kernel": 0.11255934834480286}}, "feed_forward": {"intermediate_dense": {"bias": 0.029854251071810722, "kernel": 0.18060839176177979}, "output_dense": {"bias": 0.007813147269189358, "kernel": 0.13272684812545776}}, "final_layer_norm": {"bias": 0.1974797397851944, "scale": 0.19957499206066132}, "layer_norm": {"bias": 0.14637750387191772, "scale": 0.1230667233467102}}, "5": {"attention": {"k_proj": {"bias": 1.1138767149532214e-05, "kernel": 0.040099892765283585}, "out_proj": {"bias": 0.0065377443097531796, "kernel": 0.07717025279998779}, "q_proj": {"bias": 0.008759167045354843, "kernel": 0.050802670419216156}, "v_proj": {"bias": 0.03756267949938774, "kernel": 0.10883606970310211}}, "feed_forward": {"intermediate_dense": {"bias": 0.028528494760394096, "kernel": 0.19774873554706573}, "output_dense": {"bias": 0.00439264765009284, "kernel": 0.11630585789680481}}, "final_layer_norm": {"bias": 0.1979135125875473, "scale": 0.21613231301307678}, "layer_norm": {"bias": 0.13786856830120087, "scale": 0.12307372689247131}}, "6": {"attention": {"k_proj": {"bias": 9.497131941316184e-06, "kernel": 0.03309299796819687}, "out_proj": {"bias": 0.0042800442315638065, "kernel": 0.05752479285001755}, "q_proj": {"bias": 0.005952971056103706, "kernel": 0.03820596635341644}, "v_proj": {"bias": 0.019765272736549377, "kernel": 0.0700731873512268}}, "feed_forward": {"intermediate_dense": {"bias": 0.014015424996614456, "kernel": 0.10993222892284393}, "output_dense": {"bias": 0.003552541835233569, "kernel": 0.06198006868362427}}, "final_layer_norm": {"bias": 0.08279841393232346, "scale": 0.1033489927649498}, "layer_norm": {"bias": 0.08875390887260437, "scale": 0.08801350742578506}}, "7": {"attention": {"k_proj": {"bias": 8.305349183501676e-06, "kernel": 0.02493010275065899}, "out_proj": {"bias": 0.0032034695614129305, "kernel": 0.03852525353431702}, "q_proj": {"bias": 0.004448656924068928, "kernel": 0.029889430850744247}, "v_proj": {"bias": 0.01669190265238285, "kernel": 0.04915356636047363}}, "feed_forward": {"intermediate_dense": {"bias": 0.009438328444957733, "kernel": 0.06788736581802368}, "output_dense": {"bias": 0.001637768349610269, "kernel": 0.028960686177015305}}, "final_layer_norm": {"bias": 0.10429926961660385, "scale": 0.12397060543298721}, "layer_norm": {"bias": 0.06768227368593216, "scale": 0.06890910863876343}}, "8": {"attention": {"k_proj": {"bias": 3.624029659476946e-06, "kernel": 0.010071169584989548}, "out_proj": {"bias": 0.0016006645746529102, "kernel": 0.01920405961573124}, "q_proj": {"bias": 0.0015132054686546326, "kernel": 0.009457355365157127}, "v_proj": {"bias": 0.006836294662207365, "kernel": 0.02054772526025772}}, "feed_forward": {"intermediate_dense": {"bias": 0.001426941016688943, "kernel": 0.025523120537400246}, "output_dense": {"bias": 0.0014155321987345815, "kernel": 0.022529449313879013}}, "final_layer_norm": {"bias": 0.012654873542487621, "scale": 0.017398104071617126}, "layer_norm": {"bias": 0.03265748918056488, "scale": 0.03671739995479584}}, "9": {"attention": {"k_proj": {"bias": 1.60206559485232e-06, "kernel": 0.005215200129896402}, "out_proj": {"bias": 0.001441936125047505, "kernel": 0.019099462777376175}, "q_proj": {"bias": 0.0010555217741057277, "kernel": 0.00537400646135211}, "v_proj": {"bias": 0.00635305792093277, "kernel": 0.017087457701563835}}, "feed_forward": {"intermediate_dense": {"bias": 0.0005063117714598775, "kernel": 0.01536525972187519}, "output_dense": {"bias": 0.00139874336309731, "kernel": 0.01615232415497303}}, "final_layer_norm": {"bias": 0.0022344086319208145, "scale": 0.0028617442585527897}, "layer_norm": {"bias": 0.02930845320224762, "scale": 0.018842769786715508}}}, "pos_conv_embed": {"conv": {"bias": 0.10097971558570862, "weight_g": 0.08083780854940414, "weight_v": 0.5826525688171387}}}, "feature_extractor": {"conv_layers": {"0": {"conv": {"bias": 0.0, "kernel": 0.0}, "layer_norm": {"bias": 0.0, "scale": 0.0}}, "1": {"conv": {"bias": 0.0, "kernel": 0.0}, "layer_norm": {"bias": 0.0, "scale": 0.0}}, "2": {"conv": {"bias": 0.0, "kernel": 0.0}, "layer_norm": {"bias": 0.0, "scale": 0.0}}, "3": {"conv": {"bias": 0.0, "kernel": 0.0}, "layer_norm": {"bias": 0.0, "scale": 0.0}}, "4": {"conv": {"bias": 0.0, "kernel": 0.0}, "layer_norm": {"bias": 0.0, "scale": 0.0}}, "5": {"conv": {"bias": 0.0, "kernel": 0.0}, "layer_norm": {"bias": 0.0, "scale": 0.0}}, "6": {"conv": {"bias": 0.0, "kernel": 0.0}, "layer_norm": {"bias": 0.0, "scale": 0.0}}}}, "feature_projection": {"layer_norm": {"bias": 0.20916703343391418, "scale": 0.15897098183631897}, "projection": {"bias": 0.04994938522577286, "kernel": 1.07721745967865}}, "masked_spec_embed": 0.0}}, "layer_param_norm/": {"decoder": {"model": {"decoder": {"embed_positions": {"embedding": 58.11711502075195}, "embed_tokens": {"embedding": 614.154052734375}, "layernorm_embedding": {"bias": 2.4326412677764893, "scale": 14.106523513793945}, "layers": {"0": {"encoder_attn": {"k_proj": {"bias": 1.3436226844787598, "kernel": 94.5048828125}, "out_proj": {"bias": 2.979982852935791, "kernel": 73.4836196899414}, "q_proj": {"bias": 4.843000888824463, "kernel": 96.2011947631836}, "v_proj": {"bias": 1.3543726205825806, "kernel": 73.22048950195312}}, "encoder_attn_layer_norm": {"bias": 4.235548496246338, "scale": 14.604206085205078}, "fc1": {"bias": 8.584725379943848, "kernel": 107.6702880859375}, "fc2": {"bias": 2.5613932609558105, "kernel": 90.311767578125}, "final_layer_norm": {"bias": 1.4952219724655151, "scale": 12.361992835998535}, "self_attn": {"k_proj": {"bias": 1.1865863800048828, "kernel": 90.33148956298828}, "out_proj": {"bias": 2.2230091094970703, "kernel": 36.40948486328125}, "q_proj": {"bias": 18.898283004760742, "kernel": 86.76622009277344}, "v_proj": {"bias": 2.0896196365356445, "kernel": 35.76298141479492}}, "self_attn_layer_norm": {"bias": 4.977694034576416, "scale": 23.70896339416504}}, "1": {"encoder_attn": {"k_proj": {"bias": 0.9428327083587646, "kernel": 93.44721984863281}, "out_proj": {"bias": 1.8153433799743652, "kernel": 66.90109252929688}, "q_proj": {"bias": 8.71115779876709, "kernel": 95.03624725341797}, "v_proj": {"bias": 1.412581443786621, "kernel": 68.51860046386719}}, "encoder_attn_layer_norm": {"bias": 3.407933235168457, "scale": 15.122153282165527}, "fc1": {"bias": 8.16239070892334, "kernel": 108.68303680419922}, "fc2": {"bias": 2.051272392272949, "kernel": 78.74549102783203}, "final_layer_norm": {"bias": 1.2968448400497437, "scale": 11.608719825744629}, "self_attn": {"k_proj": {"bias": 2.5194456577301025, "kernel": 83.59282684326172}, "out_proj": {"bias": 1.6754883527755737, "kernel": 38.96104049682617}, "q_proj": {"bias": 8.410796165466309, "kernel": 83.80982208251953}, "v_proj": {"bias": 0.935066819190979, "kernel": 39.35820770263672}}, "self_attn_layer_norm": {"bias": 3.5405585765838623, "scale": 25.78156280517578}}, "10": {"encoder_attn": {"k_proj": {"bias": 10.58764934539795, "kernel": 92.59211730957031}, "out_proj": {"bias": 2.537656307220459, "kernel": 61.145103454589844}, "q_proj": {"bias": 4.925878524780273, "kernel": 94.61139678955078}, "v_proj": {"bias": 1.2230350971221924, "kernel": 62.20232391357422}}, "encoder_attn_layer_norm": {"bias": 2.108487129211426, "scale": 16.940673828125}, "fc1": {"bias": 6.348001003265381, "kernel": 86.46847534179688}, "fc2": {"bias": 2.054652690887451, "kernel": 62.26055908203125}, "final_layer_norm": {"bias": 0.7569279670715332, "scale": 13.047319412231445}, "self_attn": {"k_proj": {"bias": 7.426407814025879, "kernel": 78.1574935913086}, "out_proj": {"bias": 0.5733494758605957, "kernel": 38.61314010620117}, "q_proj": {"bias": 8.276473999023438, "kernel": 79.49523162841797}, "v_proj": {"bias": 0.2637999951839447, "kernel": 41.38172149658203}}, "self_attn_layer_norm": {"bias": 2.1998722553253174, "scale": 24.147558212280273}}, "11": {"encoder_attn": {"k_proj": {"bias": 44.2944450378418, "kernel": 92.008544921875}, "out_proj": {"bias": 2.1107113361358643, "kernel": 69.27115631103516}, "q_proj": {"bias": 6.052280426025391, "kernel": 96.63239288330078}, "v_proj": {"bias": 0.8517287969589233, "kernel": 71.89142608642578}}, "encoder_attn_layer_norm": {"bias": 4.453308582305908, "scale": 17.13739013671875}, "fc1": {"bias": 4.4321184158325195, "kernel": 81.14264678955078}, "fc2": {"bias": 0.8952088356018066, "kernel": 46.79474639892578}, "final_layer_norm": {"bias": 2.470918655395508, "scale": 48.319419860839844}, "self_attn": {"k_proj": {"bias": 58.056396484375, "kernel": 81.23486328125}, "out_proj": {"bias": 0.6313948631286621, "kernel": 37.580406188964844}, "q_proj": {"bias": 10.843825340270996, "kernel": 85.051513671875}, "v_proj": {"bias": 0.16369317471981049, "kernel": 35.28926086425781}}, "self_attn_layer_norm": {"bias": 2.5757603645324707, "scale": 20.21198272705078}}, "2": {"encoder_attn": {"k_proj": {"bias": 1.36637282371521, "kernel": 91.34264373779297}, "out_proj": {"bias": 1.6449049711227417, "kernel": 68.31389617919922}, "q_proj": {"bias": 6.352076053619385, "kernel": 93.5359878540039}, "v_proj": {"bias": 1.2951818704605103, "kernel": 71.76292419433594}}, "encoder_attn_layer_norm": {"bias": 2.905444383621216, "scale": 15.695459365844727}, "fc1": {"bias": 8.01304817199707, "kernel": 106.57068634033203}, "fc2": {"bias": 2.090430974960327, "kernel": 74.14126586914062}, "final_layer_norm": {"bias": 1.4443482160568237, "scale": 11.891447067260742}, "self_attn": {"k_proj": {"bias": 1.639186143875122, "kernel": 80.40776062011719}, "out_proj": {"bias": 1.1713868379592896, "kernel": 41.034645080566406}, "q_proj": {"bias": 8.928056716918945, "kernel": 79.54090118408203}, "v_proj": {"bias": 0.5743991732597351, "kernel": 43.63880157470703}}, "self_attn_layer_norm": {"bias": 2.5566253662109375, "scale": 22.152864456176758}}, "3": {"encoder_attn": {"k_proj": {"bias": 1.468528151512146, "kernel": 91.38245391845703}, "out_proj": {"bias": 1.3379414081573486, "kernel": 64.7074966430664}, "q_proj": {"bias": 6.288889408111572, "kernel": 93.5470962524414}, "v_proj": {"bias": 1.0908479690551758, "kernel": 67.07949829101562}}, "encoder_attn_layer_norm": {"bias": 2.647975444793701, "scale": 15.231871604919434}, "fc1": {"bias": 7.2708539962768555, "kernel": 103.407470703125}, "fc2": {"bias": 2.4473917484283447, "kernel": 75.25543975830078}, "final_layer_norm": {"bias": 0.913488507270813, "scale": 12.227962493896484}, "self_attn": {"k_proj": {"bias": 2.397507905960083, "kernel": 78.11325073242188}, "out_proj": {"bias": 1.0595494508743286, "kernel": 43.838504791259766}, "q_proj": {"bias": 6.982239246368408, "kernel": 78.897705078125}, "v_proj": {"bias": 0.7266519665718079, "kernel": 46.74177551269531}}, "self_attn_layer_norm": {"bias": 1.7357004880905151, "scale": 24.207630157470703}}, "4": {"encoder_attn": {"k_proj": {"bias": 2.0629429817199707, "kernel": 95.31353759765625}, "out_proj": {"bias": 1.2406041622161865, "kernel": 67.83168029785156}, "q_proj": {"bias": 5.678467750549316, "kernel": 96.58917999267578}, "v_proj": {"bias": 0.7568002343177795, "kernel": 69.63357543945312}}, "encoder_attn_layer_norm": {"bias": 2.291189193725586, "scale": 15.504029273986816}, "fc1": {"bias": 6.789959907531738, "kernel": 98.11848449707031}, "fc2": {"bias": 2.6845154762268066, "kernel": 74.25277709960938}, "final_layer_norm": {"bias": 0.5983816981315613, "scale": 12.782552719116211}, "self_attn": {"k_proj": {"bias": 1.8189221620559692, "kernel": 76.42529296875}, "out_proj": {"bias": 0.9529430866241455, "kernel": 43.576499938964844}, "q_proj": {"bias": 6.992264270782471, "kernel": 77.15050506591797}, "v_proj": {"bias": 0.4205707311630249, "kernel": 47.40883255004883}}, "self_attn_layer_norm": {"bias": 1.5938223600387573, "scale": 23.9239501953125}}, "5": {"encoder_attn": {"k_proj": {"bias": 3.6890664100646973, "kernel": 95.5152816772461}, "out_proj": {"bias": 0.9384456276893616, "kernel": 65.50871276855469}, "q_proj": {"bias": 3.9954452514648438, "kernel": 97.28777313232422}, "v_proj": {"bias": 0.6282899975776672, "kernel": 64.92552185058594}}, "encoder_attn_layer_norm": {"bias": 2.186138868331909, "scale": 15.336478233337402}, "fc1": {"bias": 6.947918891906738, "kernel": 97.2540512084961}, "fc2": {"bias": 2.3825466632843018, "kernel": 73.37749481201172}, "final_layer_norm": {"bias": 0.7313886880874634, "scale": 12.739490509033203}, "self_attn": {"k_proj": {"bias": 2.2293927669525146, "kernel": 77.18207550048828}, "out_proj": {"bias": 0.8972251415252686, "kernel": 39.702545166015625}, "q_proj": {"bias": 6.831601142883301, "kernel": 78.41766357421875}, "v_proj": {"bias": 0.3210265338420868, "kernel": 43.100616455078125}}, "self_attn_layer_norm": {"bias": 1.5320919752120972, "scale": 25.55414390563965}}, "6": {"encoder_attn": {"k_proj": {"bias": 2.84926438331604, "kernel": 98.051025390625}, "out_proj": {"bias": 1.0777729749679565, "kernel": 66.45055389404297}, "q_proj": {"bias": 5.35632848739624, "kernel": 100.40483856201172}, "v_proj": {"bias": 0.7619112730026245, "kernel": 66.1573257446289}}, "encoder_attn_layer_norm": {"bias": 2.1359341144561768, "scale": 15.91808795928955}, "fc1": {"bias": 6.857090473175049, "kernel": 95.23121643066406}, "fc2": {"bias": 2.439351797103882, "kernel": 65.83731079101562}, "final_layer_norm": {"bias": 0.6829109787940979, "scale": 13.160418510437012}, "self_attn": {"k_proj": {"bias": 1.8109793663024902, "kernel": 77.92413330078125}, "out_proj": {"bias": 0.7429792881011963, "kernel": 37.971946716308594}, "q_proj": {"bias": 5.835843086242676, "kernel": 79.34928131103516}, "v_proj": {"bias": 0.21429745852947235, "kernel": 41.66492462158203}}, "self_attn_layer_norm": {"bias": 1.7612333297729492, "scale": 24.881845474243164}}, "7": {"encoder_attn": {"k_proj": {"bias": 2.7482666969299316, "kernel": 95.43659973144531}, "out_proj": {"bias": 1.380357265472412, "kernel": 62.0806884765625}, "q_proj": {"bias": 6.721969127655029, "kernel": 97.59810638427734}, "v_proj": {"bias": 0.8795636296272278, "kernel": 62.26142120361328}}, "encoder_attn_layer_norm": {"bias": 2.0428121089935303, "scale": 15.988516807556152}, "fc1": {"bias": 7.2896599769592285, "kernel": 92.345458984375}, "fc2": {"bias": 2.5348734855651855, "kernel": 64.27216339111328}, "final_layer_norm": {"bias": 0.5528438091278076, "scale": 12.9879789352417}, "self_attn": {"k_proj": {"bias": 2.5634522438049316, "kernel": 78.52615356445312}, "out_proj": {"bias": 0.7399742007255554, "kernel": 33.54991149902344}, "q_proj": {"bias": 6.200771331787109, "kernel": 80.02249145507812}, "v_proj": {"bias": 0.2545686960220337, "kernel": 36.81795883178711}}, "self_attn_layer_norm": {"bias": 1.968164086341858, "scale": 25.61897087097168}}, "8": {"encoder_attn": {"k_proj": {"bias": 5.648555278778076, "kernel": 93.64107513427734}, "out_proj": {"bias": 1.5685137510299683, "kernel": 58.93069839477539}, "q_proj": {"bias": 5.603804111480713, "kernel": 95.50863647460938}, "v_proj": {"bias": 1.201375126838684, "kernel": 60.25454330444336}}, "encoder_attn_layer_norm": {"bias": 2.0115392208099365, "scale": 15.265798568725586}, "fc1": {"bias": 6.969101428985596, "kernel": 91.38919830322266}, "fc2": {"bias": 2.3548667430877686, "kernel": 64.05999755859375}, "final_layer_norm": {"bias": 0.6495308876037598, "scale": 13.045710563659668}, "self_attn": {"k_proj": {"bias": 1.8263764381408691, "kernel": 79.19358825683594}, "out_proj": {"bias": 0.7575787305831909, "kernel": 32.841556549072266}, "q_proj": {"bias": 7.271490097045898, "kernel": 81.85496520996094}, "v_proj": {"bias": 0.47302117943763733, "kernel": 36.607574462890625}}, "self_attn_layer_norm": {"bias": 1.7295911312103271, "scale": 25.65311622619629}}, "9": {"encoder_attn": {"k_proj": {"bias": 11.397842407226562, "kernel": 92.93269348144531}, "out_proj": {"bias": 1.8243788480758667, "kernel": 56.957557678222656}, "q_proj": {"bias": 5.121211051940918, "kernel": 94.7365951538086}, "v_proj": {"bias": 1.484121561050415, "kernel": 58.32129669189453}}, "encoder_attn_layer_norm": {"bias": 1.502446174621582, "scale": 16.13802719116211}, "fc1": {"bias": 7.338865756988525, "kernel": 90.27828216552734}, "fc2": {"bias": 2.0869359970092773, "kernel": 64.22966766357422}, "final_layer_norm": {"bias": 0.6612036228179932, "scale": 13.04432487487793}, "self_attn": {"k_proj": {"bias": 4.388544082641602, "kernel": 80.33892822265625}, "out_proj": {"bias": 0.6234767436981201, "kernel": 33.11110305786133}, "q_proj": {"bias": 6.9557271003723145, "kernel": 82.31918334960938}, "v_proj": {"bias": 0.3340129554271698, "kernel": 37.688148498535156}}, "self_attn_layer_norm": {"bias": 1.8461397886276245, "scale": 24.963699340820312}}}}}}, "encoder": {"adapter": {"layers": {"0": {"conv": {"bias": 0.010662875138223171, "kernel": 50.1640739440918}}, "1": {"conv": {"bias": 0.010519884526729584, "kernel": 50.171669006347656}}, "2": {"conv": {"bias": 0.01166580617427826, "kernel": 50.170719146728516}}}}, "encoder": {"layer_norm": {"bias": 0.26451829075813293, "scale": 4.800480842590332}, "layers": {"0": {"attention": {"k_proj": {"bias": 0.03494168445467949, "kernel": 77.11620330810547}, "out_proj": {"bias": 6.814887523651123, "kernel": 108.30722045898438}, "q_proj": {"bias": 14.293266296386719, "kernel": 73.85565948486328}, "v_proj": {"bias": 1.418501615524292, "kernel": 85.79047393798828}}, "feed_forward": {"intermediate_dense": {"bias": 6.723220348358154, "kernel": 282.06207275390625}, "output_dense": {"bias": 3.495791435241699, "kernel": 271.8985900878906}}, "final_layer_norm": {"bias": 2.5147225856781006, "scale": 6.266326427459717}, "layer_norm": {"bias": 1.7378885746002197, "scale": 5.840572357177734}}, "1": {"attention": {"k_proj": {"bias": 0.5125874876976013, "kernel": 104.67235565185547}, "out_proj": {"bias": 2.8598856925964355, "kernel": 110.42671203613281}, "q_proj": {"bias": 9.678413391113281, "kernel": 104.75859069824219}, "v_proj": {"bias": 1.137494444847107, "kernel": 97.5046157836914}}, "feed_forward": {"intermediate_dense": {"bias": 2.3362832069396973, "kernel": 273.8770751953125}, "output_dense": {"bias": 2.4614198207855225, "kernel": 254.66966247558594}}, "final_layer_norm": {"bias": 2.4016690254211426, "scale": 7.192074775695801}, "layer_norm": {"bias": 1.153796672821045, "scale": 6.099413871765137}}, "10": {"attention": {"k_proj": {"bias": 0.6048656105995178, "kernel": 130.8582000732422}, "out_proj": {"bias": 3.913703680038452, "kernel": 146.04844665527344}, "q_proj": {"bias": 9.123319625854492, "kernel": 130.6387176513672}, "v_proj": {"bias": 2.2406952381134033, "kernel": 147.96499633789062}}, "feed_forward": {"intermediate_dense": {"bias": 4.6886210441589355, "kernel": 313.176513671875}, "output_dense": {"bias": 6.426457405090332, "kernel": 310.5086669921875}}, "final_layer_norm": {"bias": 11.110428810119629, "scale": 37.49863815307617}, "layer_norm": {"bias": 1.0594130754470825, "scale": 6.822112083435059}}, "11": {"attention": {"k_proj": {"bias": 1.0851818323135376, "kernel": 130.6464080810547}, "out_proj": {"bias": 4.055002689361572, "kernel": 151.5869903564453}, "q_proj": {"bias": 9.334198951721191, "kernel": 129.83460998535156}, "v_proj": {"bias": 2.3762176036834717, "kernel": 154.13523864746094}}, "feed_forward": {"intermediate_dense": {"bias": 6.092442512512207, "kernel": 311.75592041015625}, "output_dense": {"bias": 9.868654251098633, "kernel": 313.9928283691406}}, "final_layer_norm": {"bias": 12.55875301361084, "scale": 45.600303649902344}, "layer_norm": {"bias": 1.085222840309143, "scale": 7.346126079559326}}, "12": {"attention": {"k_proj": {"bias": 1.868220329284668, "kernel": 129.7105255126953}, "out_proj": {"bias": 5.1156535148620605, "kernel": 151.6835479736328}, "q_proj": {"bias": 8.650224685668945, "kernel": 129.0090789794922}, "v_proj": {"bias": 3.2437045574188232, "kernel": 154.69039916992188}}, "feed_forward": {"intermediate_dense": {"bias": 4.572490215301514, "kernel": 295.4952697753906}, "output_dense": {"bias": 6.202803611755371, "kernel": 282.0585632324219}}, "final_layer_norm": {"bias": 6.555354595184326, "scale": 18.1469783782959}, "layer_norm": {"bias": 1.5738931894302368, "scale": 8.01997184753418}}, "13": {"attention": {"k_proj": {"bias": 1.612104892730713, "kernel": 122.41880798339844}, "out_proj": {"bias": 3.1937506198883057, "kernel": 152.11424255371094}, "q_proj": {"bias": 9.267769813537598, "kernel": 121.65782165527344}, "v_proj": {"bias": 2.6480135917663574, "kernel": 152.53521728515625}}, "feed_forward": {"intermediate_dense": {"bias": 4.859166145324707, "kernel": 287.12322998046875}, "output_dense": {"bias": 4.87112283706665, "kernel": 281.0758056640625}}, "final_layer_norm": {"bias": 7.385702133178711, "scale": 22.75259780883789}, "layer_norm": {"bias": 1.572802186012268, "scale": 8.175799369812012}}, "14": {"attention": {"k_proj": {"bias": 0.9815863966941833, "kernel": 126.48646545410156}, "out_proj": {"bias": 3.9507412910461426, "kernel": 150.54299926757812}, "q_proj": {"bias": 8.501983642578125, "kernel": 124.47171783447266}, "v_proj": {"bias": 2.6173415184020996, "kernel": 147.41798400878906}}, "feed_forward": {"intermediate_dense": {"bias": 5.027058124542236, "kernel": 282.82049560546875}, "output_dense": {"bias": 5.374758720397949, "kernel": 275.1122131347656}}, "final_layer_norm": {"bias": 6.4135613441467285, "scale": 20.74764633178711}, "layer_norm": {"bias": 1.5298587083816528, "scale": 8.570819854736328}}, "15": {"attention": {"k_proj": {"bias": 1.8426316976547241, "kernel": 123.35442352294922}, "out_proj": {"bias": 3.5583090782165527, "kernel": 155.22744750976562}, "q_proj": {"bias": 8.42642879486084, "kernel": 121.00657653808594}, "v_proj": {"bias": 3.7885055541992188, "kernel": 153.47296142578125}}, "feed_forward": {"intermediate_dense": {"bias": 5.54867696762085, "kernel": 281.9224853515625}, "output_dense": {"bias": 4.283926010131836, "kernel": 271.753173828125}}, "final_layer_norm": {"bias": 4.9235711097717285, "scale": 18.906570434570312}, "layer_norm": {"bias": 1.4994099140167236, "scale": 8.726380348205566}}, "16": {"attention": {"k_proj": {"bias": 2.4644510746002197, "kernel": 111.25455474853516}, "out_proj": {"bias": 3.3641183376312256, "kernel": 160.1565704345703}, "q_proj": {"bias": 8.987558364868164, "kernel": 107.38188171386719}, "v_proj": {"bias": 3.2992258071899414, "kernel": 162.05767822265625}}, "feed_forward": {"intermediate_dense": {"bias": 5.620339393615723, "kernel": 283.26763916015625}, "output_dense": {"bias": 4.438972473144531, "kernel": 267.02825927734375}}, "final_layer_norm": {"bias": 5.432890892028809, "scale": 22.532976150512695}, "layer_norm": {"bias": 1.441110610961914, "scale": 9.484601974487305}}, "17": {"attention": {"k_proj": {"bias": 2.7954020500183105, "kernel": 113.37714385986328}, "out_proj": {"bias": 3.863956928253174, "kernel": 164.38290405273438}, "q_proj": {"bias": 8.34001636505127, "kernel": 109.3065185546875}, "v_proj": {"bias": 3.9232370853424072, "kernel": 164.2174835205078}}, "feed_forward": {"intermediate_dense": {"bias": 6.767458438873291, "kernel": 279.1385192871094}, "output_dense": {"bias": 3.5571823120117188, "kernel": 267.31964111328125}}, "final_layer_norm": {"bias": 3.980884313583374, "scale": 20.268936157226562}, "layer_norm": {"bias": 1.4181233644485474, "scale": 10.14484691619873}}, "18": {"attention": {"k_proj": {"bias": 5.470330238342285, "kernel": 108.28589630126953}, "out_proj": {"bias": 3.2715563774108887, "kernel": 161.3241729736328}, "q_proj": {"bias": 7.251437664031982, "kernel": 104.44432067871094}, "v_proj": {"bias": 4.183315753936768, "kernel": 162.61961364746094}}, "feed_forward": {"intermediate_dense": {"bias": 5.609297275543213, "kernel": 281.8549499511719}, "output_dense": {"bias": 4.394558429718018, "kernel": 257.8499755859375}}, "final_layer_norm": {"bias": 5.987644672393799, "scale": 30.829397201538086}, "layer_norm": {"bias": 1.597630500793457, "scale": 10.41945743560791}}, "19": {"attention": {"k_proj": {"bias": 7.467156410217285, "kernel": 102.64733123779297}, "out_proj": {"bias": 3.2516911029815674, "kernel": 167.23304748535156}, "q_proj": {"bias": 6.482592582702637, "kernel": 97.6107177734375}, "v_proj": {"bias": 4.305199146270752, "kernel": 167.6278076171875}}, "feed_forward": {"intermediate_dense": {"bias": 6.156229496002197, "kernel": 272.5770263671875}, "output_dense": {"bias": 3.8195691108703613, "kernel": 253.28399658203125}}, "final_layer_norm": {"bias": 6.152090072631836, "scale": 34.08858108520508}, "layer_norm": {"bias": 1.797480821609497, "scale": 11.417618751525879}}, "2": {"attention": {"k_proj": {"bias": 0.4069085717201233, "kernel": 110.21129608154297}, "out_proj": {"bias": 2.069960832595825, "kernel": 114.20024108886719}, "q_proj": {"bias": 8.665688514709473, "kernel": 109.20825958251953}, "v_proj": {"bias": 0.9563086032867432, "kernel": 106.45647430419922}}, "feed_forward": {"intermediate_dense": {"bias": 2.3022682666778564, "kernel": 280.53759765625}, "output_dense": {"bias": 2.0167582035064697, "kernel": 260.7087707519531}}, "final_layer_norm": {"bias": 2.3594143390655518, "scale": 7.303609848022461}, "layer_norm": {"bias": 1.0205110311508179, "scale": 6.955028533935547}}, "20": {"attention": {"k_proj": {"bias": 4.711760997772217, "kernel": 88.6990966796875}, "out_proj": {"bias": 2.8294260501861572, "kernel": 175.88510131835938}, "q_proj": {"bias": 4.3511433601379395, "kernel": 83.57088470458984}, "v_proj": {"bias": 5.2271246910095215, "kernel": 175.32728576660156}}, "feed_forward": {"intermediate_dense": {"bias": 6.189459323883057, "kernel": 262.8561706542969}, "output_dense": {"bias": 3.2152481079101562, "kernel": 245.58824157714844}}, "final_layer_norm": {"bias": 6.5343098640441895, "scale": 42.77993392944336}, "layer_norm": {"bias": 2.3866829872131348, "scale": 13.592259407043457}}, "21": {"attention": {"k_proj": {"bias": 8.614314079284668, "kernel": 77.62751007080078}, "out_proj": {"bias": 3.2871389389038086, "kernel": 168.9432830810547}, "q_proj": {"bias": 3.2974941730499268, "kernel": 74.38258361816406}, "v_proj": {"bias": 4.8117828369140625, "kernel": 167.14938354492188}}, "feed_forward": {"intermediate_dense": {"bias": 5.911465644836426, "kernel": 244.37989807128906}, "output_dense": {"bias": 1.4526047706604004, "kernel": 233.3193817138672}}, "final_layer_norm": {"bias": 7.746399879455566, "scale": 55.23497009277344}, "layer_norm": {"bias": 2.890018939971924, "scale": 15.653545379638672}}, "22": {"attention": {"k_proj": {"bias": 11.001053810119629, "kernel": 70.23458099365234}, "out_proj": {"bias": 1.9036134481430054, "kernel": 138.97097778320312}, "q_proj": {"bias": 2.714299440383911, "kernel": 70.1622314453125}, "v_proj": {"bias": 5.095273017883301, "kernel": 134.72250366210938}}, "feed_forward": {"intermediate_dense": {"bias": 6.085084915161133, "kernel": 234.28016662597656}, "output_dense": {"bias": 1.9195820093154907, "kernel": 215.61317443847656}}, "final_layer_norm": {"bias": 12.323036193847656, "scale": 61.20780944824219}, "layer_norm": {"bias": 2.348065137863159, "scale": 15.208549499511719}}, "23": {"attention": {"k_proj": {"bias": 6.555962085723877, "kernel": 78.00003814697266}, "out_proj": {"bias": 2.0626087188720703, "kernel": 140.8603515625}, "q_proj": {"bias": 5.786411285400391, "kernel": 73.50723266601562}, "v_proj": {"bias": 6.02974796295166, "kernel": 143.897705078125}}, "feed_forward": {"intermediate_dense": {"bias": 6.217076778411865, "kernel": 201.0343017578125}, "output_dense": {"bias": 0.7252612709999084, "kernel": 125.60209655761719}}, "final_layer_norm": {"bias": 4.734806537628174, "scale": 27.71601676940918}, "layer_norm": {"bias": 1.0875698328018188, "scale": 11.55836009979248}}, "3": {"attention": {"k_proj": {"bias": 0.18326698243618011, "kernel": 119.25263977050781}, "out_proj": {"bias": 1.8820536136627197, "kernel": 116.35952758789062}, "q_proj": {"bias": 7.361374855041504, "kernel": 120.10768127441406}, "v_proj": {"bias": 0.810678243637085, "kernel": 111.52812194824219}}, "feed_forward": {"intermediate_dense": {"bias": 2.0838372707366943, "kernel": 283.2486877441406}, "output_dense": {"bias": 2.051933765411377, "kernel": 259.9745178222656}}, "final_layer_norm": {"bias": 2.2448692321777344, "scale": 7.258777618408203}, "layer_norm": {"bias": 0.8867139220237732, "scale": 7.426033973693848}}, "4": {"attention": {"k_proj": {"bias": 0.7118229866027832, "kernel": 124.03813171386719}, "out_proj": {"bias": 1.6172691583633423, "kernel": 121.10663604736328}, "q_proj": {"bias": 8.4681396484375, "kernel": 122.57655334472656}, "v_proj": {"bias": 1.1929394006729126, "kernel": 118.79817199707031}}, "feed_forward": {"intermediate_dense": {"bias": 1.9473854303359985, "kernel": 284.6288146972656}, "output_dense": {"bias": 1.6239995956420898, "kernel": 257.7264709472656}}, "final_layer_norm": {"bias": 2.100123167037964, "scale": 7.200130939483643}, "layer_norm": {"bias": 1.1937754154205322, "scale": 6.84559965133667}}, "5": {"attention": {"k_proj": {"bias": 0.6982775330543518, "kernel": 120.28306579589844}, "out_proj": {"bias": 1.8659156560897827, "kernel": 129.3507843017578}, "q_proj": {"bias": 8.441648483276367, "kernel": 119.09827423095703}, "v_proj": {"bias": 1.2326970100402832, "kernel": 129.14305114746094}}, "feed_forward": {"intermediate_dense": {"bias": 2.1792142391204834, "kernel": 280.68682861328125}, "output_dense": {"bias": 1.8617591857910156, "kernel": 257.16162109375}}, "final_layer_norm": {"bias": 2.161855459213257, "scale": 7.165393829345703}, "layer_norm": {"bias": 1.23024320602417, "scale": 6.433177947998047}}, "6": {"attention": {"k_proj": {"bias": 0.2670230269432068, "kernel": 123.5765380859375}, "out_proj": {"bias": 1.6782166957855225, "kernel": 120.1172866821289}, "q_proj": {"bias": 7.17927885055542, "kernel": 123.54830932617188}, "v_proj": {"bias": 1.1905819177627563, "kernel": 119.64702606201172}}, "feed_forward": {"intermediate_dense": {"bias": 3.1326568126678467, "kernel": 283.4269714355469}, "output_dense": {"bias": 1.9350062608718872, "kernel": 260.4270324707031}}, "final_layer_norm": {"bias": 2.7130236625671387, "scale": 8.390787124633789}, "layer_norm": {"bias": 0.9011468291282654, "scale": 7.117128849029541}}, "7": {"attention": {"k_proj": {"bias": 0.45881396532058716, "kernel": 117.64019012451172}, "out_proj": {"bias": 2.535649538040161, "kernel": 128.38597106933594}, "q_proj": {"bias": 8.063240051269531, "kernel": 117.92418670654297}, "v_proj": {"bias": 1.5216361284255981, "kernel": 126.03490447998047}}, "feed_forward": {"intermediate_dense": {"bias": 3.059353828430176, "kernel": 278.994384765625}, "output_dense": {"bias": 2.2597203254699707, "kernel": 270.811767578125}}, "final_layer_norm": {"bias": 2.4217488765716553, "scale": 7.401240825653076}, "layer_norm": {"bias": 1.0410571098327637, "scale": 6.691290378570557}}, "8": {"attention": {"k_proj": {"bias": 0.7014403343200684, "kernel": 121.97848510742188}, "out_proj": {"bias": 3.4911699295043945, "kernel": 135.91360473632812}, "q_proj": {"bias": 9.189191818237305, "kernel": 122.31185150146484}, "v_proj": {"bias": 1.8781499862670898, "kernel": 134.98971557617188}}, "feed_forward": {"intermediate_dense": {"bias": 2.7755894660949707, "kernel": 300.2033996582031}, "output_dense": {"bias": 3.840137243270874, "kernel": 287.60052490234375}}, "final_layer_norm": {"bias": 6.93441915512085, "scale": 20.86276626586914}, "layer_norm": {"bias": 0.9869874119758606, "scale": 6.73464822769165}}, "9": {"attention": {"k_proj": {"bias": 0.5955453515052795, "kernel": 121.50640106201172}, "out_proj": {"bias": 4.259273052215576, "kernel": 140.76930236816406}, "q_proj": {"bias": 9.543315887451172, "kernel": 122.70530700683594}, "v_proj": {"bias": 1.9614927768707275, "kernel": 136.73973083496094}}, "feed_forward": {"intermediate_dense": {"bias": 6.232731342315674, "kernel": 307.1314392089844}, "output_dense": {"bias": 6.060487747192383, "kernel": 305.42010498046875}}, "final_layer_norm": {"bias": 10.532364845275879, "scale": 39.37934494018555}, "layer_norm": {"bias": 0.9431830048561096, "scale": 6.89916467666626}}}, "pos_conv_embed": {"conv": {"bias": 15.41067123413086, "weight_g": 21.01583480834961, "weight_v": 215.22080993652344}}}, "feature_extractor": {"conv_layers": {"0": {"conv": {"bias": 0.5982058644294739, "kernel": 8.08896541595459}, "layer_norm": {"bias": 10.069783210754395, "scale": 10.451257705688477}}, "1": {"conv": {"bias": 4.74075174331665, "kernel": 90.8435287475586}, "layer_norm": {"bias": 6.922820091247559, "scale": 19.5467586517334}}, "2": {"conv": {"bias": 6.7732415199279785, "kernel": 146.13897705078125}, "layer_norm": {"bias": 9.044225692749023, "scale": 19.424888610839844}}, "3": {"conv": {"bias": 5.224758148193359, "kernel": 159.10508728027344}, "layer_norm": {"bias": 8.319666862487793, "scale": 17.64743423461914}}, "4": {"conv": {"bias": 4.434978008270264, "kernel": 157.35813903808594}, "layer_norm": {"bias": 9.193974494934082, "scale": 15.562357902526855}}, "5": {"conv": {"bias": 5.297643661499023, "kernel": 131.1835174560547}, "layer_norm": {"bias": 10.735219955444336, "scale": 13.812533378601074}}, "6": {"conv": {"bias": 5.615579128265381, "kernel": 136.41822814941406}, "layer_norm": {"bias": 12.515308380126953, "scale": 11.152680397033691}}}}, "feature_projection": {"layer_norm": {"bias": 10.161255836486816, "scale": 27.72752571105957}, "projection": {"bias": 4.438388824462891, "kernel": 88.27880859375}}, "masked_spec_embed": 26.247730255126953}}, "train/learning_rate": 3.0000002880115062e-05, "train/loss": 4.60268497467041, "train/param_norm": 2534.2275390625, "_timestamp": 1649194129, "_runtime": 1506, "_step": 166, "eval/loss": 4.616994857788086, "eval/wer": 1.0104942772898766, "predictions/epoch_1": {"_type": "table-file", "path": "media/table/predictions/epoch_1_166_cd7954b6d932424bb480.table.json", "sha256": "cd7954b6d932424bb480eca029caeb05868c5b91820a6d26dbf7a4c54dea4dc8", "size": 10457, "artifact_path": "wandb-client-artifact://12mubouboytts7pbt7wnu6ny9yhwfcvdvp8d78zw30974syhfj1jcekrgrnan9c5z8iuvp45pigrbyyh6gw0vl3j9lt1rwpncgyadzuehoeqg6sn3b9xmc078cmgg1kq:latest/predictions/epoch_1.table.json", "_latest_artifact_path": "wandb-client-artifact://12mubouboytts7pbt7wnu6ny9yhwfcvdvp8d78zw30974syhfj1jcekrgrnan9c5z8iuvp45pigrbyyh6gw0vl3j9lt1rwpncgyadzuehoeqg6sn3b9xmc078cmgg1kq:latest/predictions/epoch_1.table.json", "ncols": 2, "nrows": 50}}
\ No newline at end of file
+{"train/decoder_grad_norm": 6.969270706176758, "train/decoder_param_norm": 1043.1859130859375, "train/encoder_grad_norm": 1.4057550430297852, "train/encoder_param_norm": 2309.41845703125, "train/grad_norm": 7.109632968902588, "layer_grad_norm/": {"decoder": {"model": {"decoder": {"embed_positions": {"embedding": 0.5117830634117126}, "embed_tokens": {"embedding": 1.1998742818832397}, "layernorm_embedding": {"bias": 0.2228890061378479, "scale": 0.15787996351718903}, "layers": {"0": {"encoder_attn": {"k_proj": {"bias": 1.1884704690601211e-05, "kernel": 0.010384045541286469}, "out_proj": {"bias": 0.07051730901002884, "kernel": 0.10179866850376129}, "q_proj": {"bias": 0.0005757305189035833, "kernel": 0.012427772395312786}, "v_proj": {"bias": 0.17104488611221313, "kernel": 0.041392866522073746}}, "encoder_attn_layer_norm": {"bias": 0.10716868191957474, "scale": 0.10110125690698624}, "fc1": {"bias": 0.04969731718301773, "kernel": 0.7886783480644226}, "fc2": {"bias": 0.08068706840276718, "kernel": 0.8774661421775818}, "final_layer_norm": {"bias": 0.231051966547966, "scale": 0.1301460713148117}, "self_attn": {"k_proj": {"bias": 3.018476490979083e-05, "kernel": 0.15450438857078552}, "out_proj": {"bias": 0.1174507662653923, "kernel": 0.7987715601921082}, "q_proj": {"bias": 0.014772974886000156, "kernel": 0.21989428997039795}, "v_proj": {"bias": 0.13174845278263092, "kernel": 0.8497021198272705}}, "self_attn_layer_norm": {"bias": 0.06730392575263977, "scale": 0.06256284564733505}}, "1": {"encoder_attn": {"k_proj": {"bias": 5.782705557066947e-06, "kernel": 0.004698821809142828}, "out_proj": {"bias": 0.06683740019798279, "kernel": 0.09665905684232712}, "q_proj": {"bias": 0.0001470006100134924, "kernel": 0.006146976724267006}, "v_proj": {"bias": 0.1466769576072693, "kernel": 0.03513868898153305}}, "encoder_attn_layer_norm": {"bias": 0.10971161723136902, "scale": 0.10633502155542374}, "fc1": {"bias": 0.05027240142226219, "kernel": 0.7868102192878723}, "fc2": {"bias": 0.08194814622402191, "kernel": 0.8334754109382629}, "final_layer_norm": {"bias": 0.2724083364009857, "scale": 0.13667769730091095}, "self_attn": {"k_proj": {"bias": 1.9709097614395432e-05, "kernel": 0.11050131916999817}, "out_proj": {"bias": 0.13011232018470764, "kernel": 0.4560726284980774}, "q_proj": {"bias": 0.011685396544635296, "kernel": 0.14933884143829346}, "v_proj": {"bias": 0.14886687695980072, "kernel": 0.5208406448364258}}, "self_attn_layer_norm": {"bias": 0.06399120390415192, "scale": 0.0597182996571064}}, "10": {"encoder_attn": {"k_proj": {"bias": 4.237310349708423e-06, "kernel": 0.0033278081100434065}, "out_proj": {"bias": 0.09429243952035904, "kernel": 0.12043017894029617}, "q_proj": {"bias": 0.00010958474013023078, "kernel": 0.0033953727688640356}, "v_proj": {"bias": 0.183123379945755, "kernel": 0.044296350330114365}}, "encoder_attn_layer_norm": {"bias": 0.15246617794036865, "scale": 0.13445161283016205}, "fc1": {"bias": 0.06822530180215836, "kernel": 2.838428020477295}, "fc2": {"bias": 0.14114779233932495, "kernel": 1.2782968282699585}, "final_layer_norm": {"bias": 1.3923081159591675, "scale": 0.2813446819782257}, "self_attn": {"k_proj": {"bias": 1.4987470422056504e-05, "kernel": 0.09808193147182465}, "out_proj": {"bias": 0.33817312121391296, "kernel": 0.3301580548286438}, "q_proj": {"bias": 0.007489969953894615, "kernel": 0.10408112406730652}, "v_proj": {"bias": 0.37949079275131226, "kernel": 0.9953370690345764}}, "self_attn_layer_norm": {"bias": 0.09078381210565567, "scale": 0.08854929357767105}}, "11": {"encoder_attn": {"k_proj": {"bias": 6.327686151053058e-06, "kernel": 0.006727192550897598}, "out_proj": {"bias": 0.12685038149356842, "kernel": 0.12352851778268814}, "q_proj": {"bias": 0.0003102236951235682, "kernel": 0.0059141297824680805}, "v_proj": {"bias": 0.2638256847858429, "kernel": 0.06254562735557556}}, "encoder_attn_layer_norm": {"bias": 0.15296398103237152, "scale": 0.13720858097076416}, "fc1": {"bias": 0.06340250372886658, "kernel": 2.406972646713257}, "fc2": {"bias": 0.15614399313926697, "kernel": 1.4546396732330322}, "final_layer_norm": {"bias": 0.12907014787197113, "scale": 0.05075053125619888}, "self_attn": {"k_proj": {"bias": 1.3952081644674763e-05, "kernel": 0.07334674149751663}, "out_proj": {"bias": 0.8997290730476379, "kernel": 0.37400856614112854}, "q_proj": {"bias": 0.004955133888870478, "kernel": 0.08000137656927109}, "v_proj": {"bias": 0.8556796312332153, "kernel": 0.6879818439483643}}, "self_attn_layer_norm": {"bias": 0.12080614268779755, "scale": 0.17146295309066772}}, "2": {"encoder_attn": {"k_proj": {"bias": 4.554362931230571e-06, "kernel": 0.0052012912929058075}, "out_proj": {"bias": 0.07746843248605728, "kernel": 0.10529642552137375}, "q_proj": {"bias": 0.00026940429233945906, "kernel": 0.005280209239572287}, "v_proj": {"bias": 0.17219410836696625, "kernel": 0.03998562693595886}}, "encoder_attn_layer_norm": {"bias": 0.10770189762115479, "scale": 0.11770331859588623}, "fc1": {"bias": 0.04484928771853447, "kernel": 0.7265751361846924}, "fc2": {"bias": 0.0879487544298172, "kernel": 0.8496302366256714}, "final_layer_norm": {"bias": 0.2788235545158386, "scale": 0.14027826488018036}, "self_attn": {"k_proj": {"bias": 1.867869468696881e-05, "kernel": 0.10047563910484314}, "out_proj": {"bias": 0.13781775534152985, "kernel": 0.37114372849464417}, "q_proj": {"bias": 0.010185548104345798, "kernel": 0.12143644690513611}, "v_proj": {"bias": 0.176111102104187, "kernel": 0.48070481419563293}}, "self_attn_layer_norm": {"bias": 0.07551655918359756, "scale": 0.07633473724126816}}, "3": {"encoder_attn": {"k_proj": {"bias": 3.75643185179797e-06, "kernel": 0.004228970501571894}, "out_proj": {"bias": 0.0740702748298645, "kernel": 0.08248641341924667}, "q_proj": {"bias": 0.00021324875706341118, "kernel": 0.0038996038492769003}, "v_proj": {"bias": 0.15333180129528046, "kernel": 0.03647708147764206}}, "encoder_attn_layer_norm": {"bias": 0.11628811061382294, "scale": 0.11092574149370193}, "fc1": {"bias": 0.04610592871904373, "kernel": 0.7018353939056396}, "fc2": {"bias": 0.09111737459897995, "kernel": 0.8374897837638855}, "final_layer_norm": {"bias": 0.2850073575973511, "scale": 0.1575641632080078}, "self_attn": {"k_proj": {"bias": 1.4689080671814736e-05, "kernel": 0.10236077755689621}, "out_proj": {"bias": 0.1434292495250702, "kernel": 0.25565409660339355}, "q_proj": {"bias": 0.010950295254588127, "kernel": 0.12489160895347595}, "v_proj": {"bias": 0.17960533499717712, "kernel": 0.40027931332588196}}, "self_attn_layer_norm": {"bias": 0.07180269807577133, "scale": 0.07309168577194214}}, "4": {"encoder_attn": {"k_proj": {"bias": 4.58631666333531e-06, "kernel": 0.0045629823580384254}, "out_proj": {"bias": 0.08055750280618668, "kernel": 0.07053619623184204}, "q_proj": {"bias": 0.00020153954392299056, "kernel": 0.004369232803583145}, "v_proj": {"bias": 0.1700308918952942, "kernel": 0.03961788862943649}}, "encoder_attn_layer_norm": {"bias": 0.11302211135625839, "scale": 0.10816352069377899}, "fc1": {"bias": 0.04346245527267456, "kernel": 0.6954390406608582}, "fc2": {"bias": 0.09931685775518417, "kernel": 0.8692843914031982}, "final_layer_norm": {"bias": 0.3319553732872009, "scale": 0.1503801941871643}, "self_attn": {"k_proj": {"bias": 1.5018401427369099e-05, "kernel": 0.12500926852226257}, "out_proj": {"bias": 0.1531442403793335, "kernel": 0.29529309272766113}, "q_proj": {"bias": 0.01281694695353508, "kernel": 0.143911212682724}, "v_proj": {"bias": 0.19408272206783295, "kernel": 0.3585151135921478}}, "self_attn_layer_norm": {"bias": 0.07813063263893127, "scale": 0.07336484640836716}}, "5": {"encoder_attn": {"k_proj": {"bias": 3.818025561486138e-06, "kernel": 0.003916461952030659}, "out_proj": {"bias": 0.07940594106912613, "kernel": 0.0589476116001606}, "q_proj": {"bias": 0.00017110994667746127, "kernel": 0.004327178001403809}, "v_proj": {"bias": 0.17055687308311462, "kernel": 0.039873156696558}}, "encoder_attn_layer_norm": {"bias": 0.11560037732124329, "scale": 0.11377429962158203}, "fc1": {"bias": 0.04432540386915207, "kernel": 0.6914775967597961}, "fc2": {"bias": 0.10287611931562424, "kernel": 0.8692248463630676}, "final_layer_norm": {"bias": 0.32355692982673645, "scale": 0.16169960796833038}, "self_attn": {"k_proj": {"bias": 2.449178100505378e-05, "kernel": 0.1160995364189148}, "out_proj": {"bias": 0.1528330147266388, "kernel": 0.3025151193141937}, "q_proj": {"bias": 0.011123402044177055, "kernel": 0.13983044028282166}, "v_proj": {"bias": 0.1960059106349945, "kernel": 0.3581959307193756}}, "self_attn_layer_norm": {"bias": 0.07529620081186295, "scale": 0.07391269505023956}}, "6": {"encoder_attn": {"k_proj": {"bias": 5.171521934244083e-06, "kernel": 0.005456449463963509}, "out_proj": {"bias": 0.0809267982840538, "kernel": 0.07032608240842819}, "q_proj": {"bias": 0.00021908830967731774, "kernel": 0.004780640359967947}, "v_proj": {"bias": 0.16742560267448425, "kernel": 0.039506688714027405}}, "encoder_attn_layer_norm": {"bias": 0.12282489985227585, "scale": 0.11739695817232132}, "fc1": {"bias": 0.04502282664179802, "kernel": 0.7490335702896118}, "fc2": {"bias": 0.10666675865650177, "kernel": 0.8993347883224487}, "final_layer_norm": {"bias": 0.2752576172351837, "scale": 0.1449299156665802}, "self_attn": {"k_proj": {"bias": 2.519337613193784e-05, "kernel": 0.13132350146770477}, "out_proj": {"bias": 0.15100647509098053, "kernel": 0.3694905638694763}, "q_proj": {"bias": 0.011680039577186108, "kernel": 0.15310031175613403}, "v_proj": {"bias": 0.19045744836330414, "kernel": 0.42738011479377747}}, "self_attn_layer_norm": {"bias": 0.07840970903635025, "scale": 0.07609746605157852}}, "7": {"encoder_attn": {"k_proj": {"bias": 4.146530500293011e-06, "kernel": 0.004527896177023649}, "out_proj": {"bias": 0.08013690263032913, "kernel": 0.07447285205125809}, "q_proj": {"bias": 0.00019355042604729533, "kernel": 0.00412247097119689}, "v_proj": {"bias": 0.15657877922058105, "kernel": 0.03628624975681305}}, "encoder_attn_layer_norm": {"bias": 0.12783850729465485, "scale": 0.11700034886598587}, "fc1": {"bias": 0.04964722320437431, "kernel": 0.9330015778541565}, "fc2": {"bias": 0.10850771516561508, "kernel": 1.000636339187622}, "final_layer_norm": {"bias": 0.2531479597091675, "scale": 0.2975310981273651}, "self_attn": {"k_proj": {"bias": 1.9447033992037177e-05, "kernel": 0.11471318453550339}, "out_proj": {"bias": 0.15105275809764862, "kernel": 0.36407604813575745}, "q_proj": {"bias": 0.011037531308829784, "kernel": 0.13717085123062134}, "v_proj": {"bias": 0.15301616489887238, "kernel": 0.458305299282074}}, "self_attn_layer_norm": {"bias": 0.07678239792585373, "scale": 0.0748312920331955}}, "8": {"encoder_attn": {"k_proj": {"bias": 3.346356834299513e-06, "kernel": 0.0029839801136404276}, "out_proj": {"bias": 0.07559487968683243, "kernel": 0.0946722999215126}, "q_proj": {"bias": 9.915108239511028e-05, "kernel": 0.003328086342662573}, "v_proj": {"bias": 0.13498057425022125, "kernel": 0.03205062076449394}}, "encoder_attn_layer_norm": {"bias": 0.12227523326873779, "scale": 0.11871527135372162}, "fc1": {"bias": 0.05124358832836151, "kernel": 0.8128637671470642}, "fc2": {"bias": 0.1054784283041954, "kernel": 1.0374855995178223}, "final_layer_norm": {"bias": 0.23821581900119781, "scale": 0.14554870128631592}, "self_attn": {"k_proj": {"bias": 1.7555121303303167e-05, "kernel": 0.11432763934135437}, "out_proj": {"bias": 0.13978871703147888, "kernel": 0.3838534653186798}, "q_proj": {"bias": 0.010160624049603939, "kernel": 0.1265949010848999}, "v_proj": {"bias": 0.14518855512142181, "kernel": 0.615729808807373}}, "self_attn_layer_norm": {"bias": 0.07356501370668411, "scale": 0.07273858040571213}}, "9": {"encoder_attn": {"k_proj": {"bias": 4.0308063944394235e-06, "kernel": 0.0041751377284526825}, "out_proj": {"bias": 0.07924921065568924, "kernel": 0.11959613114595413}, "q_proj": {"bias": 0.00018713920144364238, "kernel": 0.0043449969962239265}, "v_proj": {"bias": 0.13957169651985168, "kernel": 0.03352203965187073}}, "encoder_attn_layer_norm": {"bias": 0.12556833028793335, "scale": 0.12256702780723572}, "fc1": {"bias": 0.055720254778862, "kernel": 1.2716578245162964}, "fc2": {"bias": 0.112123504281044, "kernel": 1.132912516593933}, "final_layer_norm": {"bias": 0.4874364137649536, "scale": 0.27216944098472595}, "self_attn": {"k_proj": {"bias": 1.5074373550305609e-05, "kernel": 0.09932490438222885}, "out_proj": {"bias": 0.16004310548305511, "kernel": 0.3056156635284424}, "q_proj": {"bias": 0.008054476231336594, "kernel": 0.10338465124368668}, "v_proj": {"bias": 0.16322575509548187, "kernel": 0.7667742371559143}}, "self_attn_layer_norm": {"bias": 0.07533666491508484, "scale": 0.07303927838802338}}}}}}, "encoder": {"adapter": {"layers": {"0": {"conv": {"bias": 0.21679997444152832, "kernel": 0.6702502369880676}}, "1": {"conv": {"bias": 0.40302562713623047, "kernel": 0.5145807862281799}}, "2": {"conv": {"bias": 0.7309064269065857, "kernel": 0.48062774538993835}}}}, "encoder": {"layer_norm": {"bias": 0.24289771914482117, "scale": 0.19713689386844635}, "layers": {"0": {"attention": {"k_proj": {"bias": 5.473609235195909e-06, "kernel": 0.012446227483451366}, "out_proj": {"bias": 0.0064920992590487, "kernel": 0.06046414375305176}, "q_proj": {"bias": 0.002947595901787281, "kernel": 0.01037699356675148}, "v_proj": {"bias": 0.033787377178668976, "kernel": 0.05555132403969765}}, "feed_forward": {"intermediate_dense": {"bias": 0.01048416830599308, "kernel": 0.055988576263189316}, "output_dense": {"bias": 0.002735169604420662, "kernel": 0.045134905725717545}}, "final_layer_norm": {"bias": 0.06150330975651741, "scale": 0.07400664687156677}, "layer_norm": {"bias": 0.08231285214424133, "scale": 0.12461329996585846}}, "1": {"attention": {"k_proj": {"bias": 4.62690422864398e-06, "kernel": 0.0098298704251647}, "out_proj": {"bias": 0.0030902379658073187, "kernel": 0.038921844214200974}, "q_proj": {"bias": 0.001854123780503869, "kernel": 0.009497118182480335}, "v_proj": {"bias": 0.012128067202866077, "kernel": 0.03811175003647804}}, "feed_forward": {"intermediate_dense": {"bias": 0.0064764306880533695, "kernel": 0.03552646189928055}, "output_dense": {"bias": 0.002369693713262677, "kernel": 0.026128407567739487}}, "final_layer_norm": {"bias": 0.03657805919647217, "scale": 0.030950654298067093}, "layer_norm": {"bias": 0.04948360100388527, "scale": 0.032011643052101135}}, "10": {"attention": {"k_proj": {"bias": 1.3978476545162266e-06, "kernel": 0.005798615515232086}, "out_proj": {"bias": 0.0015868833288550377, "kernel": 0.0234896931797266}, "q_proj": {"bias": 0.0016091542784124613, "kernel": 0.006437975447624922}, "v_proj": {"bias": 0.007461432833224535, "kernel": 0.024328578263521194}}, "feed_forward": {"intermediate_dense": {"bias": 0.0008370384457521141, "kernel": 0.019723763689398766}, "output_dense": {"bias": 0.0013029554393142462, "kernel": 0.01740993931889534}}, "final_layer_norm": {"bias": 0.004321891348809004, "scale": 0.0031387759372591972}, "layer_norm": {"bias": 0.0371471531689167, "scale": 0.03303028270602226}}, "11": {"attention": {"k_proj": {"bias": 1.331522298642085e-06, "kernel": 0.006322280503809452}, "out_proj": {"bias": 0.001275384332984686, "kernel": 0.02978055737912655}, "q_proj": {"bias": 0.0014152699150145054, "kernel": 0.006734328810125589}, "v_proj": {"bias": 0.005075549706816673, "kernel": 0.021520409733057022}}, "feed_forward": {"intermediate_dense": {"bias": 0.000432295462815091, "kernel": 0.01451073121279478}, "output_dense": {"bias": 0.0012552303960546851, "kernel": 0.015911055728793144}}, "final_layer_norm": {"bias": 0.0017107807798311114, "scale": 0.0015347120352089405}, "layer_norm": {"bias": 0.02594156377017498, "scale": 0.020046798512339592}}, "12": {"attention": {"k_proj": {"bias": 1.7764855329005513e-06, "kernel": 0.005532258655875921}, "out_proj": {"bias": 0.0011224579066038132, "kernel": 0.027212003245949745}, "q_proj": {"bias": 0.0008849767618812621, "kernel": 0.00580595713108778}, "v_proj": {"bias": 0.0049216956831514835, "kernel": 0.02385624311864376}}, "feed_forward": {"intermediate_dense": {"bias": 0.0019450337858870625, "kernel": 0.02675033174455166}, "output_dense": {"bias": 0.0011028049048036337, "kernel": 0.021677428856492043}}, "final_layer_norm": {"bias": 0.007728468626737595, "scale": 0.011072246357798576}, "layer_norm": {"bias": 0.023660574108362198, "scale": 0.01693642884492874}}, "13": {"attention": {"k_proj": {"bias": 1.17454089831881e-06, "kernel": 0.00508676515892148}, "out_proj": {"bias": 0.001058893627487123, "kernel": 0.0261110607534647}, "q_proj": {"bias": 0.000804621260613203, "kernel": 0.005890473257750273}, "v_proj": {"bias": 0.00466560572385788, "kernel": 0.023298826068639755}}, "feed_forward": {"intermediate_dense": {"bias": 0.0015144250355660915, "kernel": 0.02727758325636387}, "output_dense": {"bias": 0.0009065094636753201, "kernel": 0.02077588625252247}}, "final_layer_norm": {"bias": 0.0075051747262477875, "scale": 0.01087986584752798}, "layer_norm": {"bias": 0.023178771138191223, "scale": 0.02436891756951809}}, "14": {"attention": {"k_proj": {"bias": 1.529968244540214e-06, "kernel": 0.005575601942837238}, "out_proj": {"bias": 0.0008673755219206214, "kernel": 0.01968420296907425}, "q_proj": {"bias": 0.0008510977495461702, "kernel": 0.005965071264654398}, "v_proj": {"bias": 0.0035861791111528873, "kernel": 0.018254714086651802}}, "feed_forward": {"intermediate_dense": {"bias": 0.0014124573208391666, "kernel": 0.02255995012819767}, "output_dense": {"bias": 0.0008412961033172905, "kernel": 0.015012514777481556}}, "final_layer_norm": {"bias": 0.00624546455219388, "scale": 0.006261321250349283}, "layer_norm": {"bias": 0.017770880833268166, "scale": 0.01359514705836773}}, "15": {"attention": {"k_proj": {"bias": 1.523301421002543e-06, "kernel": 0.008141463622450829}, "out_proj": {"bias": 0.000768792990129441, "kernel": 0.020340770483016968}, "q_proj": {"bias": 0.0010627361480146646, "kernel": 0.008440445177257061}, "v_proj": {"bias": 0.003393010701984167, "kernel": 0.018189558759331703}}, "feed_forward": {"intermediate_dense": {"bias": 0.001493017072789371, "kernel": 0.021281136199831963}, "output_dense": {"bias": 0.0007060510688461363, "kernel": 0.017788082361221313}}, "final_layer_norm": {"bias": 0.0070274448953568935, "scale": 0.01119965873658657}, "layer_norm": {"bias": 0.020354028791189194, "scale": 0.015358109958469868}}, "16": {"attention": {"k_proj": {"bias": 1.4672046972918906e-06, "kernel": 0.006524242926388979}, "out_proj": {"bias": 0.0006828226614743471, "kernel": 0.02142046019434929}, "q_proj": {"bias": 0.0009362387354485691, "kernel": 0.00732450233772397}, "v_proj": {"bias": 0.0029004383832216263, "kernel": 0.016688412055373192}}, "feed_forward": {"intermediate_dense": {"bias": 0.0012145887594670057, "kernel": 0.0219045951962471}, "output_dense": {"bias": 0.0006580579793080688, "kernel": 0.020583467558026314}}, "final_layer_norm": {"bias": 0.005356708075851202, "scale": 0.0069047738797962666}, "layer_norm": {"bias": 0.016287008300423622, "scale": 0.012492732144892216}}, "17": {"attention": {"k_proj": {"bias": 1.2948949006386101e-06, "kernel": 0.006665787193924189}, "out_proj": {"bias": 0.0006345617584884167, "kernel": 0.021538930013775826}, "q_proj": {"bias": 0.0008286732481792569, "kernel": 0.0069121187552809715}, "v_proj": {"bias": 0.0025541428476572037, "kernel": 0.016102001070976257}}, "feed_forward": {"intermediate_dense": {"bias": 0.0017689007800072432, "kernel": 0.03051549568772316}, "output_dense": {"bias": 0.0005231120740063488, "kernel": 0.023845301941037178}}, "final_layer_norm": {"bias": 0.008580866269767284, "scale": 0.01467498205602169}, "layer_norm": {"bias": 0.014641078189015388, "scale": 0.010583329945802689}}, "18": {"attention": {"k_proj": {"bias": 1.4464804962699418e-06, "kernel": 0.0093308100476861}, "out_proj": {"bias": 0.0004914177698083222, "kernel": 0.01819584146142006}, "q_proj": {"bias": 0.0011983370641246438, "kernel": 0.010149878449738026}, "v_proj": {"bias": 0.0024688320700079203, "kernel": 0.016792036592960358}}, "feed_forward": {"intermediate_dense": {"bias": 0.0009198479820042849, "kernel": 0.023795250803232193}, "output_dense": {"bias": 0.00043442542664706707, "kernel": 0.02401123009622097}}, "final_layer_norm": {"bias": 0.004441787488758564, "scale": 0.0062178936786949635}, "layer_norm": {"bias": 0.016286905854940414, "scale": 0.014550870284438133}}, "19": {"attention": {"k_proj": {"bias": 1.3731305443798192e-06, "kernel": 0.011217473074793816}, "out_proj": {"bias": 0.00043212290620431304, "kernel": 0.018191879615187645}, "q_proj": {"bias": 0.0012673117453232408, "kernel": 0.011789560317993164}, "v_proj": {"bias": 0.00213794712908566, "kernel": 0.0154916662722826}}, "feed_forward": {"intermediate_dense": {"bias": 0.000986509840004146, "kernel": 0.028885388746857643}, "output_dense": {"bias": 0.00036284225643612444, "kernel": 0.02663830853998661}}, "final_layer_norm": {"bias": 0.004911407362669706, "scale": 0.009552892297506332}, "layer_norm": {"bias": 0.013291558250784874, "scale": 0.010668172501027584}}, "2": {"attention": {"k_proj": {"bias": 3.8377979763026815e-06, "kernel": 0.009901314973831177}, "out_proj": {"bias": 0.0024876536335796118, "kernel": 0.029377911239862442}, "q_proj": {"bias": 0.0019141703378409147, "kernel": 0.010465304367244244}, "v_proj": {"bias": 0.010124980472028255, "kernel": 0.03445031866431236}}, "feed_forward": {"intermediate_dense": {"bias": 0.005793231073766947, "kernel": 0.03215382620692253}, "output_dense": {"bias": 0.0021973017137497663, "kernel": 0.02444424107670784}}, "final_layer_norm": {"bias": 0.027568859979510307, "scale": 0.026417221873998642}, "layer_norm": {"bias": 0.040012311190366745, "scale": 0.034563686698675156}}, "20": {"attention": {"k_proj": {"bias": 1.248317403224064e-06, "kernel": 0.017442084848880768}, "out_proj": {"bias": 0.00033371051540598273, "kernel": 0.017458025366067886}, "q_proj": {"bias": 0.0015706189442425966, "kernel": 0.016797281801700592}, "v_proj": {"bias": 0.001977138454094529, "kernel": 0.014997645281255245}}, "feed_forward": {"intermediate_dense": {"bias": 0.0008223969489336014, "kernel": 0.030204998329281807}, "output_dense": {"bias": 0.00032897639903239906, "kernel": 0.03247291222214699}}, "final_layer_norm": {"bias": 0.004012326709926128, "scale": 0.006121456157416105}, "layer_norm": {"bias": 0.014743871055543423, "scale": 0.015442288480699062}}, "21": {"attention": {"k_proj": {"bias": 7.711757348261017e-07, "kernel": 0.017593875527381897}, "out_proj": {"bias": 0.00024054452660493553, "kernel": 0.015012494288384914}, "q_proj": {"bias": 0.0012941056629642844, "kernel": 0.017408333718776703}, "v_proj": {"bias": 0.0015631928108632565, "kernel": 0.014115679077804089}}, "feed_forward": {"intermediate_dense": {"bias": 0.0007619658135809004, "kernel": 0.035559844225645065}, "output_dense": {"bias": 0.0002678300370462239, "kernel": 0.04254584014415741}}, "final_layer_norm": {"bias": 0.0032283600885421038, "scale": 0.003060641000047326}, "layer_norm": {"bias": 0.012259122915565968, "scale": 0.015054778195917606}}, "22": {"attention": {"k_proj": {"bias": 1.6521696011295717e-07, "kernel": 0.010658740065991879}, "out_proj": {"bias": 0.0002742362266872078, "kernel": 0.048993028700351715}, "q_proj": {"bias": 0.0009089713566936553, "kernel": 0.01134306751191616}, "v_proj": {"bias": 0.0016765004256740212, "kernel": 0.0418640561401844}}, "feed_forward": {"intermediate_dense": {"bias": 0.0010514555033296347, "kernel": 0.06577882915735245}, "output_dense": {"bias": 0.00039716323954053223, "kernel": 0.08095555752515793}}, "final_layer_norm": {"bias": 0.004826333839446306, "scale": 0.004279589746147394}, "layer_norm": {"bias": 0.011670369654893875, "scale": 0.017876500263810158}}, "23": {"attention": {"k_proj": {"bias": 2.342537470667594e-07, "kernel": 0.005412762518972158}, "out_proj": {"bias": 0.0004161088145337999, "kernel": 0.03890999034047127}, "q_proj": {"bias": 0.000538219406735152, "kernel": 0.0045631034299731255}, "v_proj": {"bias": 0.0033254821319133043, "kernel": 0.04772743582725525}}, "feed_forward": {"intermediate_dense": {"bias": 0.0012235299218446016, "kernel": 0.02907288633286953}, "output_dense": {"bias": 0.00042490524356253445, "kernel": 0.017651060596108437}}, "final_layer_norm": {"bias": 0.004505955148488283, "scale": 0.0041798185557127}, "layer_norm": {"bias": 0.03569646552205086, "scale": 0.061676107347011566}}, "3": {"attention": {"k_proj": {"bias": 4.920991614199011e-06, "kernel": 0.009988185949623585}, "out_proj": {"bias": 0.002168482169508934, "kernel": 0.02653035707771778}, "q_proj": {"bias": 0.0017990904161706567, "kernel": 0.010570628568530083}, "v_proj": {"bias": 0.009546158835291862, "kernel": 0.031211627647280693}}, "feed_forward": {"intermediate_dense": {"bias": 0.004873727913945913, "kernel": 0.02812979929149151}, "output_dense": {"bias": 0.001906346413306892, "kernel": 0.0211043618619442}}, "final_layer_norm": {"bias": 0.026796428486704826, "scale": 0.03137560933828354}, "layer_norm": {"bias": 0.040544264018535614, "scale": 0.03252830356359482}}, "4": {"attention": {"k_proj": {"bias": 4.02763453166699e-06, "kernel": 0.010139317251741886}, "out_proj": {"bias": 0.0019477236783131957, "kernel": 0.025585072115063667}, "q_proj": {"bias": 0.001995032886043191, "kernel": 0.01082521677017212}, "v_proj": {"bias": 0.008343137800693512, "kernel": 0.03185305371880531}}, "feed_forward": {"intermediate_dense": {"bias": 0.004344860557466745, "kernel": 0.02428542822599411}, "output_dense": {"bias": 0.0018117844592779875, "kernel": 0.02082073874771595}}, "final_layer_norm": {"bias": 0.022104766219854355, "scale": 0.02592417038977146}, "layer_norm": {"bias": 0.036661550402641296, "scale": 0.031193993985652924}}, "5": {"attention": {"k_proj": {"bias": 3.697698502946878e-06, "kernel": 0.009518596343696117}, "out_proj": {"bias": 0.0018933300161734223, "kernel": 0.022700000554323196}, "q_proj": {"bias": 0.0020964201539754868, "kernel": 0.010278989560902119}, "v_proj": {"bias": 0.008586655370891094, "kernel": 0.02735404670238495}}, "feed_forward": {"intermediate_dense": {"bias": 0.004610993433743715, "kernel": 0.024403687566518784}, "output_dense": {"bias": 0.0017975035589188337, "kernel": 0.02111838012933731}}, "final_layer_norm": {"bias": 0.028011379763484, "scale": 0.02634856477379799}, "layer_norm": {"bias": 0.03346652910113335, "scale": 0.03310469910502434}}, "6": {"attention": {"k_proj": {"bias": 2.665294232429005e-06, "kernel": 0.010031267069280148}, "out_proj": {"bias": 0.0017973947105929255, "kernel": 0.023828642442822456}, "q_proj": {"bias": 0.001745773246511817, "kernel": 0.009465263225138187}, "v_proj": {"bias": 0.006916787941008806, "kernel": 0.024542126804590225}}, "feed_forward": {"intermediate_dense": {"bias": 0.003298386698588729, "kernel": 0.020811425521969795}, "output_dense": {"bias": 0.0016820260789245367, "kernel": 0.016634657979011536}}, "final_layer_norm": {"bias": 0.017593560740351677, "scale": 0.018315790221095085}, "layer_norm": {"bias": 0.03193330764770508, "scale": 0.02766535058617592}}, "7": {"attention": {"k_proj": {"bias": 3.401724370633019e-06, "kernel": 0.008768780156970024}, "out_proj": {"bias": 0.0017190906219184399, "kernel": 0.02113587222993374}, "q_proj": {"bias": 0.0018833335489034653, "kernel": 0.009678788483142853}, "v_proj": {"bias": 0.007216728292405605, "kernel": 0.023428581655025482}}, "feed_forward": {"intermediate_dense": {"bias": 0.0037369427736848593, "kernel": 0.02015567384660244}, "output_dense": {"bias": 0.001784463762305677, "kernel": 0.019439566880464554}}, "final_layer_norm": {"bias": 0.018053365871310234, "scale": 0.019001543521881104}, "layer_norm": {"bias": 0.0280039943754673, "scale": 0.027825787663459778}}, "8": {"attention": {"k_proj": {"bias": 1.618576447981468e-06, "kernel": 0.005275240167975426}, "out_proj": {"bias": 0.002092965180054307, "kernel": 0.02469806931912899}, "q_proj": {"bias": 0.0013126700650900602, "kernel": 0.005936095025390387}, "v_proj": {"bias": 0.007994838058948517, "kernel": 0.023945240303874016}}, "feed_forward": {"intermediate_dense": {"bias": 0.0012219561031088233, "kernel": 0.016970057040452957}, "output_dense": {"bias": 0.002014836762100458, "kernel": 0.015015801414847374}}, "final_layer_norm": {"bias": 0.0061976113356649876, "scale": 0.004672948736697435}, "layer_norm": {"bias": 0.03915019705891609, "scale": 0.025009669363498688}}, "9": {"attention": {"k_proj": {"bias": 1.9515234725986375e-06, "kernel": 0.005608465988188982}, "out_proj": {"bias": 0.00196436676196754, "kernel": 0.02712547779083252}, "q_proj": {"bias": 0.001459087012335658, "kernel": 0.005978711880743504}, "v_proj": {"bias": 0.0083112558349967, "kernel": 0.02428523264825344}}, "feed_forward": {"intermediate_dense": {"bias": 0.000780800764914602, "kernel": 0.019370149821043015}, "output_dense": {"bias": 0.0018814976792782545, "kernel": 0.016260433942079544}}, "final_layer_norm": {"bias": 0.005255761556327343, "scale": 0.005534108728170395}, "layer_norm": {"bias": 0.038162991404533386, "scale": 0.028217297047376633}}}, "pos_conv_embed": {"conv": {"bias": 0.006938824895769358, "weight_g": 0.005688216537237167, "weight_v": 0.06998559087514877}}}, "feature_extractor": {"conv_layers": {"0": {"conv": {"bias": 0.0, "kernel": 0.0}, "layer_norm": {"bias": 0.0, "scale": 0.0}}, "1": {"conv": {"bias": 0.0, "kernel": 0.0}, "layer_norm": {"bias": 0.0, "scale": 0.0}}, "2": {"conv": {"bias": 0.0, "kernel": 0.0}, "layer_norm": {"bias": 0.0, "scale": 0.0}}, "3": {"conv": {"bias": 0.0, "kernel": 0.0}, "layer_norm": {"bias": 0.0, "scale": 0.0}}, "4": {"conv": {"bias": 0.0, "kernel": 0.0}, "layer_norm": {"bias": 0.0, "scale": 0.0}}, "5": {"conv": {"bias": 0.0, "kernel": 0.0}, "layer_norm": {"bias": 0.0, "scale": 0.0}}, "6": {"conv": {"bias": 0.0, "kernel": 0.0}, "layer_norm": {"bias": 0.0, "scale": 0.0}}}}, "feature_projection": {"layer_norm": {"bias": 0.020279817283153534, "scale": 0.022980444133281708}, "projection": {"bias": 0.005088440608233213, "kernel": 0.13218317925930023}}, "masked_spec_embed": 0.0}}, "layer_param_norm/": {"decoder": {"model": {"decoder": {"embed_positions": {"embedding": 58.11617660522461}, "embed_tokens": {"embedding": 613.4808349609375}, "layernorm_embedding": {"bias": 2.4291481971740723, "scale": 14.111303329467773}, "layers": {"0": {"encoder_attn": {"k_proj": {"bias": 1.3446208238601685, "kernel": 94.51432800292969}, "out_proj": {"bias": 2.978419780731201, "kernel": 73.48096466064453}, "q_proj": {"bias": 4.840251445770264, "kernel": 96.20978546142578}, "v_proj": {"bias": 1.3541523218154907, "kernel": 73.21913146972656}}, "encoder_attn_layer_norm": {"bias": 4.2332282066345215, "scale": 14.613801002502441}, "fc1": {"bias": 8.581632614135742, "kernel": 107.69560241699219}, "fc2": {"bias": 2.561241626739502, "kernel": 90.32441711425781}, "final_layer_norm": {"bias": 1.4953548908233643, "scale": 12.36380672454834}, "self_attn": {"k_proj": {"bias": 1.1871508359909058, "kernel": 90.33028411865234}, "out_proj": {"bias": 2.2230384349823, "kernel": 36.40009307861328}, "q_proj": {"bias": 18.887266159057617, "kernel": 86.77183532714844}, "v_proj": {"bias": 2.087671995162964, "kernel": 35.75877380371094}}, "self_attn_layer_norm": {"bias": 4.976274490356445, "scale": 23.71310043334961}}, "1": {"encoder_attn": {"k_proj": {"bias": 0.9430420398712158, "kernel": 93.45063018798828}, "out_proj": {"bias": 1.8148866891860962, "kernel": 66.89716339111328}, "q_proj": {"bias": 8.707131385803223, "kernel": 95.04048156738281}, "v_proj": {"bias": 1.4090996980667114, "kernel": 68.51573181152344}}, "encoder_attn_layer_norm": {"bias": 3.406759738922119, "scale": 15.12764835357666}, "fc1": {"bias": 8.162083625793457, "kernel": 108.70297241210938}, "fc2": {"bias": 2.050367832183838, "kernel": 78.7579116821289}, "final_layer_norm": {"bias": 1.29628324508667, "scale": 11.609668731689453}, "self_attn": {"k_proj": {"bias": 2.5190482139587402, "kernel": 83.59741973876953}, "out_proj": {"bias": 1.6753737926483154, "kernel": 38.96614074707031}, "q_proj": {"bias": 8.41357421875, "kernel": 83.8143081665039}, "v_proj": {"bias": 0.9352753758430481, "kernel": 39.36337661743164}}, "self_attn_layer_norm": {"bias": 3.539625644683838, "scale": 25.78399085998535}}, "10": {"encoder_attn": {"k_proj": {"bias": 10.586920738220215, "kernel": 92.59004974365234}, "out_proj": {"bias": 2.5359206199645996, "kernel": 61.136085510253906}, "q_proj": {"bias": 4.920466423034668, "kernel": 94.60966491699219}, "v_proj": {"bias": 1.2233223915100098, "kernel": 62.192420959472656}}, "encoder_attn_layer_norm": {"bias": 2.1128716468811035, "scale": 16.943584442138672}, "fc1": {"bias": 6.351865291595459, "kernel": 86.48519134521484}, "fc2": {"bias": 2.0547664165496826, "kernel": 62.26908874511719}, "final_layer_norm": {"bias": 0.7571423649787903, "scale": 13.04433536529541}, "self_attn": {"k_proj": {"bias": 7.421319961547852, "kernel": 78.15802001953125}, "out_proj": {"bias": 0.5738811492919922, "kernel": 38.61479187011719}, "q_proj": {"bias": 8.277215003967285, "kernel": 79.49541473388672}, "v_proj": {"bias": 0.2634337246417999, "kernel": 41.38407516479492}}, "self_attn_layer_norm": {"bias": 2.199143171310425, "scale": 24.14556312561035}}, "11": {"encoder_attn": {"k_proj": {"bias": 44.29511260986328, "kernel": 92.00979614257812}, "out_proj": {"bias": 2.108304738998413, "kernel": 69.25939178466797}, "q_proj": {"bias": 6.05100679397583, "kernel": 96.6326904296875}, "v_proj": {"bias": 0.8534229397773743, "kernel": 71.87825012207031}}, "encoder_attn_layer_norm": {"bias": 4.462810516357422, "scale": 17.141803741455078}, "fc1": {"bias": 4.436861038208008, "kernel": 81.17860412597656}, "fc2": {"bias": 0.8943515419960022, "kernel": 46.80553436279297}, "final_layer_norm": {"bias": 2.4701766967773438, "scale": 48.30855941772461}, "self_attn": {"k_proj": {"bias": 58.05742263793945, "kernel": 81.23695373535156}, "out_proj": {"bias": 0.6310449838638306, "kernel": 37.579368591308594}, "q_proj": {"bias": 10.848437309265137, "kernel": 85.05213928222656}, "v_proj": {"bias": 0.16326774656772614, "kernel": 35.28947448730469}}, "self_attn_layer_norm": {"bias": 2.5748021602630615, "scale": 20.209474563598633}}, "2": {"encoder_attn": {"k_proj": {"bias": 1.3665695190429688, "kernel": 91.34716796875}, "out_proj": {"bias": 1.6453496217727661, "kernel": 68.30916595458984}, "q_proj": {"bias": 6.350149631500244, "kernel": 93.54060363769531}, "v_proj": {"bias": 1.2921791076660156, "kernel": 71.75907897949219}}, "encoder_attn_layer_norm": {"bias": 2.9031755924224854, "scale": 15.70043659210205}, "fc1": {"bias": 8.012948036193848, "kernel": 106.58692169189453}, "fc2": {"bias": 2.0895698070526123, "kernel": 74.15347290039062}, "final_layer_norm": {"bias": 1.444828748703003, "scale": 11.891616821289062}, "self_attn": {"k_proj": {"bias": 1.639306664466858, "kernel": 80.4103012084961}, "out_proj": {"bias": 1.170893907546997, "kernel": 41.0363883972168}, "q_proj": {"bias": 8.928243637084961, "kernel": 79.54376983642578}, "v_proj": {"bias": 0.5741354823112488, "kernel": 43.64109802246094}}, "self_attn_layer_norm": {"bias": 2.5564069747924805, "scale": 22.154987335205078}}, "3": {"encoder_attn": {"k_proj": {"bias": 1.4689631462097168, "kernel": 91.38751983642578}, "out_proj": {"bias": 1.3382670879364014, "kernel": 64.70364379882812}, "q_proj": {"bias": 6.288036346435547, "kernel": 93.55248260498047}, "v_proj": {"bias": 1.0881900787353516, "kernel": 67.076416015625}}, "encoder_attn_layer_norm": {"bias": 2.6434812545776367, "scale": 15.23826789855957}, "fc1": {"bias": 7.269317150115967, "kernel": 103.4242935180664}, "fc2": {"bias": 2.4463510513305664, "kernel": 75.26640319824219}, "final_layer_norm": {"bias": 0.9131566286087036, "scale": 12.22771167755127}, "self_attn": {"k_proj": {"bias": 2.39734148979187, "kernel": 78.11389923095703}, "out_proj": {"bias": 1.0607231855392456, "kernel": 43.839752197265625}, "q_proj": {"bias": 6.983859539031982, "kernel": 78.8984146118164}, "v_proj": {"bias": 0.725636899471283, "kernel": 46.74432373046875}}, "self_attn_layer_norm": {"bias": 1.735128402709961, "scale": 24.20937728881836}}, "4": {"encoder_attn": {"k_proj": {"bias": 2.062436819076538, "kernel": 95.31460571289062}, "out_proj": {"bias": 1.2399461269378662, "kernel": 67.8276138305664}, "q_proj": {"bias": 5.67819881439209, "kernel": 96.59147644042969}, "v_proj": {"bias": 0.7556883096694946, "kernel": 69.63009643554688}}, "encoder_attn_layer_norm": {"bias": 2.290184497833252, "scale": 15.508909225463867}, "fc1": {"bias": 6.789956092834473, "kernel": 98.1388931274414}, "fc2": {"bias": 2.683797836303711, "kernel": 74.26532745361328}, "final_layer_norm": {"bias": 0.598393440246582, "scale": 12.780433654785156}, "self_attn": {"k_proj": {"bias": 1.8196377754211426, "kernel": 76.42881774902344}, "out_proj": {"bias": 0.9525681734085083, "kernel": 43.57954788208008}, "q_proj": {"bias": 6.99235725402832, "kernel": 77.15428924560547}, "v_proj": {"bias": 0.4213024079799652, "kernel": 47.41223907470703}}, "self_attn_layer_norm": {"bias": 1.5932204723358154, "scale": 23.92409324645996}}, "5": {"encoder_attn": {"k_proj": {"bias": 3.6891989707946777, "kernel": 95.51834869384766}, "out_proj": {"bias": 0.9384638071060181, "kernel": 65.50463104248047}, "q_proj": {"bias": 3.9967305660247803, "kernel": 97.2914810180664}, "v_proj": {"bias": 0.6266705393791199, "kernel": 64.9223403930664}}, "encoder_attn_layer_norm": {"bias": 2.185028076171875, "scale": 15.34260082244873}, "fc1": {"bias": 6.948307514190674, "kernel": 97.27291870117188}, "fc2": {"bias": 2.3813047409057617, "kernel": 73.387939453125}, "final_layer_norm": {"bias": 0.7318801283836365, "scale": 12.737685203552246}, "self_attn": {"k_proj": {"bias": 2.2284533977508545, "kernel": 77.18477630615234}, "out_proj": {"bias": 0.896189272403717, "kernel": 39.707000732421875}, "q_proj": {"bias": 6.830775737762451, "kernel": 78.41999816894531}, "v_proj": {"bias": 0.3216107487678528, "kernel": 43.10456848144531}}, "self_attn_layer_norm": {"bias": 1.5315375328063965, "scale": 25.553329467773438}}, "6": {"encoder_attn": {"k_proj": {"bias": 2.8478596210479736, "kernel": 98.05200958251953}, "out_proj": {"bias": 1.0782148838043213, "kernel": 66.44552612304688}, "q_proj": {"bias": 5.35598611831665, "kernel": 100.40625}, "v_proj": {"bias": 0.7591345906257629, "kernel": 66.15312957763672}}, "encoder_attn_layer_norm": {"bias": 2.136373281478882, "scale": 15.92380142211914}, "fc1": {"bias": 6.858093738555908, "kernel": 95.25308990478516}, "fc2": {"bias": 2.43931245803833, "kernel": 65.84937286376953}, "final_layer_norm": {"bias": 0.6843094229698181, "scale": 13.158320426940918}, "self_attn": {"k_proj": {"bias": 1.8098224401474, "kernel": 77.92806243896484}, "out_proj": {"bias": 0.7429025769233704, "kernel": 37.97957992553711}, "q_proj": {"bias": 5.837282180786133, "kernel": 79.35345458984375}, "v_proj": {"bias": 0.2144051194190979, "kernel": 41.6716194152832}}, "self_attn_layer_norm": {"bias": 1.7615715265274048, "scale": 24.88120460510254}}, "7": {"encoder_attn": {"k_proj": {"bias": 2.7497823238372803, "kernel": 95.43936157226562}, "out_proj": {"bias": 1.3811990022659302, "kernel": 62.07341766357422}, "q_proj": {"bias": 6.7219390869140625, "kernel": 97.60175323486328}, "v_proj": {"bias": 0.8766195178031921, "kernel": 62.25519943237305}}, "encoder_attn_layer_norm": {"bias": 2.041248083114624, "scale": 15.994450569152832}, "fc1": {"bias": 7.290560245513916, "kernel": 92.36561584472656}, "fc2": {"bias": 2.5339276790618896, "kernel": 64.28421020507812}, "final_layer_norm": {"bias": 0.5513193607330322, "scale": 12.985711097717285}, "self_attn": {"k_proj": {"bias": 2.563246488571167, "kernel": 78.52962493896484}, "out_proj": {"bias": 0.7406561970710754, "kernel": 33.55664825439453}, "q_proj": {"bias": 6.20374870300293, "kernel": 80.0257339477539}, "v_proj": {"bias": 0.25385069847106934, "kernel": 36.824317932128906}}, "self_attn_layer_norm": {"bias": 1.9684561491012573, "scale": 25.61815071105957}}, "8": {"encoder_attn": {"k_proj": {"bias": 5.648403644561768, "kernel": 93.64200592041016}, "out_proj": {"bias": 1.567598581314087, "kernel": 58.924137115478516}, "q_proj": {"bias": 5.599798202514648, "kernel": 95.51033782958984}, "v_proj": {"bias": 1.19956636428833, "kernel": 60.24892044067383}}, "encoder_attn_layer_norm": {"bias": 2.01301646232605, "scale": 15.271136283874512}, "fc1": {"bias": 6.970829963684082, "kernel": 91.40879821777344}, "fc2": {"bias": 2.3541948795318604, "kernel": 64.06977081298828}, "final_layer_norm": {"bias": 0.6478220820426941, "scale": 13.042739868164062}, "self_attn": {"k_proj": {"bias": 1.824428677558899, "kernel": 79.19727325439453}, "out_proj": {"bias": 0.7574214935302734, "kernel": 32.847965240478516}, "q_proj": {"bias": 7.271189212799072, "kernel": 81.85832214355469}, "v_proj": {"bias": 0.4727972745895386, "kernel": 36.61323165893555}}, "self_attn_layer_norm": {"bias": 1.7291041612625122, "scale": 25.651718139648438}}, "9": {"encoder_attn": {"k_proj": {"bias": 11.398046493530273, "kernel": 92.9337387084961}, "out_proj": {"bias": 1.8230020999908447, "kernel": 56.94907760620117}, "q_proj": {"bias": 5.119080543518066, "kernel": 94.73835754394531}, "v_proj": {"bias": 1.4828431606292725, "kernel": 58.3131217956543}}, "encoder_attn_layer_norm": {"bias": 1.5034377574920654, "scale": 16.142271041870117}, "fc1": {"bias": 7.342137813568115, "kernel": 90.29335021972656}, "fc2": {"bias": 2.087050676345825, "kernel": 64.23690032958984}, "final_layer_norm": {"bias": 0.6599453687667847, "scale": 13.041131019592285}, "self_attn": {"k_proj": {"bias": 4.385507583618164, "kernel": 80.33969116210938}, "out_proj": {"bias": 0.6231709718704224, "kernel": 33.116539001464844}, "q_proj": {"bias": 6.954474925994873, "kernel": 82.31964874267578}, "v_proj": {"bias": 0.33380359411239624, "kernel": 37.693572998046875}}, "self_attn_layer_norm": {"bias": 1.846030592918396, "scale": 24.96213150024414}}}}}}, "encoder": {"adapter": {"layers": {"0": {"conv": {"bias": 0.032993778586387634, "kernel": 50.15877914428711}}, "1": {"conv": {"bias": 0.03287384659051895, "kernel": 50.17110824584961}}, "2": {"conv": {"bias": 0.036346059292554855, "kernel": 50.16891098022461}}}}, "encoder": {"layer_norm": {"bias": 0.2698884904384613, "scale": 4.786179542541504}, "layers": {"0": {"attention": {"k_proj": {"bias": 0.03641233220696449, "kernel": 77.11711883544922}, "out_proj": {"bias": 6.81569766998291, "kernel": 108.3079833984375}, "q_proj": {"bias": 14.291744232177734, "kernel": 73.85765838623047}, "v_proj": {"bias": 1.4181500673294067, "kernel": 85.79084014892578}}, "feed_forward": {"intermediate_dense": {"bias": 6.7228875160217285, "kernel": 282.0617980957031}, "output_dense": {"bias": 3.4962046146392822, "kernel": 271.8996276855469}}, "final_layer_norm": {"bias": 2.514223337173462, "scale": 6.266380786895752}, "layer_norm": {"bias": 1.7374335527420044, "scale": 5.840876579284668}}, "1": {"attention": {"k_proj": {"bias": 0.51225346326828, "kernel": 104.6733169555664}, "out_proj": {"bias": 2.860507011413574, "kernel": 110.42646026611328}, "q_proj": {"bias": 9.679121017456055, "kernel": 104.75940704345703}, "v_proj": {"bias": 1.1377623081207275, "kernel": 97.50421142578125}}, "feed_forward": {"intermediate_dense": {"bias": 2.3363449573516846, "kernel": 273.87847900390625}, "output_dense": {"bias": 2.4622647762298584, "kernel": 254.66964721679688}}, "final_layer_norm": {"bias": 2.4033701419830322, "scale": 7.1916823387146}, "layer_norm": {"bias": 1.1537431478500366, "scale": 6.0988287925720215}}, "10": {"attention": {"k_proj": {"bias": 0.6055452227592468, "kernel": 130.85751342773438}, "out_proj": {"bias": 3.913083076477051, "kernel": 146.04930114746094}, "q_proj": {"bias": 9.122178077697754, "kernel": 130.63832092285156}, "v_proj": {"bias": 2.2427315711975098, "kernel": 147.96510314941406}}, "feed_forward": {"intermediate_dense": {"bias": 4.689115047454834, "kernel": 313.1783752441406}, "output_dense": {"bias": 6.426085948944092, "kernel": 310.5103759765625}}, "final_layer_norm": {"bias": 11.109566688537598, "scale": 37.4998779296875}, "layer_norm": {"bias": 1.0594775676727295, "scale": 6.820804119110107}}, "11": {"attention": {"k_proj": {"bias": 1.0849424600601196, "kernel": 130.64724731445312}, "out_proj": {"bias": 4.053185939788818, "kernel": 151.58767700195312}, "q_proj": {"bias": 9.33646011352539, "kernel": 129.8351593017578}, "v_proj": {"bias": 2.3742997646331787, "kernel": 154.13580322265625}}, "feed_forward": {"intermediate_dense": {"bias": 6.092465400695801, "kernel": 311.7571716308594}, "output_dense": {"bias": 9.866683006286621, "kernel": 313.994140625}}, "final_layer_norm": {"bias": 12.559375762939453, "scale": 45.60036849975586}, "layer_norm": {"bias": 1.084261178970337, "scale": 7.346122741699219}}, "12": {"attention": {"k_proj": {"bias": 1.8688563108444214, "kernel": 129.7127227783203}, "out_proj": {"bias": 5.11505651473999, "kernel": 151.68455505371094}, "q_proj": {"bias": 8.65080738067627, "kernel": 129.01065063476562}, "v_proj": {"bias": 3.2418670654296875, "kernel": 154.69131469726562}}, "feed_forward": {"intermediate_dense": {"bias": 4.572788238525391, "kernel": 295.49432373046875}, "output_dense": {"bias": 6.203413963317871, "kernel": 282.0594787597656}}, "final_layer_norm": {"bias": 6.561983108520508, "scale": 18.142175674438477}, "layer_norm": {"bias": 1.5751290321350098, "scale": 8.019964218139648}}, "13": {"attention": {"k_proj": {"bias": 1.611405849456787, "kernel": 122.42012786865234}, "out_proj": {"bias": 3.1939690113067627, "kernel": 152.1156005859375}, "q_proj": {"bias": 9.267165184020996, "kernel": 121.65921020507812}, "v_proj": {"bias": 2.647199869155884, "kernel": 152.53677368164062}}, "feed_forward": {"intermediate_dense": {"bias": 4.858739852905273, "kernel": 287.12628173828125}, "output_dense": {"bias": 4.870517253875732, "kernel": 281.078857421875}}, "final_layer_norm": {"bias": 7.380073547363281, "scale": 22.75577735900879}, "layer_norm": {"bias": 1.5719197988510132, "scale": 8.176578521728516}}, "14": {"attention": {"k_proj": {"bias": 0.9826491475105286, "kernel": 126.48847198486328}, "out_proj": {"bias": 3.949955463409424, "kernel": 150.54278564453125}, "q_proj": {"bias": 8.502060890197754, "kernel": 124.47361755371094}, "v_proj": {"bias": 2.61659836769104, "kernel": 147.4178009033203}}, "feed_forward": {"intermediate_dense": {"bias": 5.0277791023254395, "kernel": 282.822265625}, "output_dense": {"bias": 5.374195575714111, "kernel": 275.11346435546875}}, "final_layer_norm": {"bias": 6.415918827056885, "scale": 20.74748992919922}, "layer_norm": {"bias": 1.530942440032959, "scale": 8.57020092010498}}, "15": {"attention": {"k_proj": {"bias": 1.843149185180664, "kernel": 123.35552215576172}, "out_proj": {"bias": 3.5570156574249268, "kernel": 155.22824096679688}, "q_proj": {"bias": 8.425776481628418, "kernel": 121.00779724121094}, "v_proj": {"bias": 3.788569211959839, "kernel": 153.47402954101562}}, "feed_forward": {"intermediate_dense": {"bias": 5.548226833343506, "kernel": 281.9232482910156}, "output_dense": {"bias": 4.284555435180664, "kernel": 271.7547912597656}}, "final_layer_norm": {"bias": 4.923277378082275, "scale": 18.906490325927734}, "layer_norm": {"bias": 1.4986186027526855, "scale": 8.726862907409668}}, "16": {"attention": {"k_proj": {"bias": 2.463609218597412, "kernel": 111.25535583496094}, "out_proj": {"bias": 3.36427640914917, "kernel": 160.1575927734375}, "q_proj": {"bias": 8.986817359924316, "kernel": 107.38331604003906}, "v_proj": {"bias": 3.299212694168091, "kernel": 162.05865478515625}}, "feed_forward": {"intermediate_dense": {"bias": 5.619777679443359, "kernel": 283.2692565917969}, "output_dense": {"bias": 4.439038276672363, "kernel": 267.0294494628906}}, "final_layer_norm": {"bias": 5.432062149047852, "scale": 22.533634185791016}, "layer_norm": {"bias": 1.4409937858581543, "scale": 9.48482894897461}}, "17": {"attention": {"k_proj": {"bias": 2.795949697494507, "kernel": 113.37761688232422}, "out_proj": {"bias": 3.8633346557617188, "kernel": 164.3834228515625}, "q_proj": {"bias": 8.340582847595215, "kernel": 109.30692291259766}, "v_proj": {"bias": 3.922921895980835, "kernel": 164.21702575683594}}, "feed_forward": {"intermediate_dense": {"bias": 6.767929553985596, "kernel": 279.1396484375}, "output_dense": {"bias": 3.557150363922119, "kernel": 267.3215026855469}}, "final_layer_norm": {"bias": 3.9805805683135986, "scale": 20.268726348876953}, "layer_norm": {"bias": 1.4180067777633667, "scale": 10.143765449523926}}, "18": {"attention": {"k_proj": {"bias": 5.470475196838379, "kernel": 108.28707122802734}, "out_proj": {"bias": 3.270418167114258, "kernel": 161.32472229003906}, "q_proj": {"bias": 7.251899242401123, "kernel": 104.44509887695312}, "v_proj": {"bias": 4.181873321533203, "kernel": 162.62098693847656}}, "feed_forward": {"intermediate_dense": {"bias": 5.608319282531738, "kernel": 281.8567810058594}, "output_dense": {"bias": 4.394289970397949, "kernel": 257.8515930175781}}, "final_layer_norm": {"bias": 5.9842610359191895, "scale": 30.830718994140625}, "layer_norm": {"bias": 1.597142219543457, "scale": 10.419777870178223}}, "19": {"attention": {"k_proj": {"bias": 7.4680562019348145, "kernel": 102.64867401123047}, "out_proj": {"bias": 3.251391887664795, "kernel": 167.23373413085938}, "q_proj": {"bias": 6.482656002044678, "kernel": 97.61215209960938}, "v_proj": {"bias": 4.30551290512085, "kernel": 167.6285400390625}}, "feed_forward": {"intermediate_dense": {"bias": 6.156556129455566, "kernel": 272.57879638671875}, "output_dense": {"bias": 3.8202831745147705, "kernel": 253.28582763671875}}, "final_layer_norm": {"bias": 6.151598930358887, "scale": 34.08909606933594}, "layer_norm": {"bias": 1.7967195510864258, "scale": 11.417969703674316}}, "2": {"attention": {"k_proj": {"bias": 0.4066357910633087, "kernel": 110.21351623535156}, "out_proj": {"bias": 2.07079815864563, "kernel": 114.20150756835938}, "q_proj": {"bias": 8.665882110595703, "kernel": 109.21023559570312}, "v_proj": {"bias": 0.9564866423606873, "kernel": 106.4573745727539}}, "feed_forward": {"intermediate_dense": {"bias": 2.302002191543579, "kernel": 280.53704833984375}, "output_dense": {"bias": 2.017303466796875, "kernel": 260.7098693847656}}, "final_layer_norm": {"bias": 2.3609602451324463, "scale": 7.302507400512695}, "layer_norm": {"bias": 1.020600438117981, "scale": 6.955983638763428}}, "20": {"attention": {"k_proj": {"bias": 4.71073579788208, "kernel": 88.69828796386719}, "out_proj": {"bias": 2.8310554027557373, "kernel": 175.88629150390625}, "q_proj": {"bias": 4.351193428039551, "kernel": 83.56985473632812}, "v_proj": {"bias": 5.227483749389648, "kernel": 175.3280792236328}}, "feed_forward": {"intermediate_dense": {"bias": 6.190395832061768, "kernel": 262.85772705078125}, "output_dense": {"bias": 3.2170679569244385, "kernel": 245.5904998779297}}, "final_layer_norm": {"bias": 6.536041736602783, "scale": 42.779544830322266}, "layer_norm": {"bias": 2.3872361183166504, "scale": 13.591468811035156}}, "21": {"attention": {"k_proj": {"bias": 8.614387512207031, "kernel": 77.62586975097656}, "out_proj": {"bias": 3.2894952297210693, "kernel": 168.9524688720703}, "q_proj": {"bias": 3.300947666168213, "kernel": 74.38176727294922}, "v_proj": {"bias": 4.814317226409912, "kernel": 167.15338134765625}}, "feed_forward": {"intermediate_dense": {"bias": 5.909729480743408, "kernel": 244.3831329345703}, "output_dense": {"bias": 1.4514439105987549, "kernel": 233.32215881347656}}, "final_layer_norm": {"bias": 7.74083948135376, "scale": 55.23556137084961}, "layer_norm": {"bias": 2.8901915550231934, "scale": 15.653864860534668}}, "22": {"attention": {"k_proj": {"bias": 11.000996589660645, "kernel": 70.23917388916016}, "out_proj": {"bias": 1.9078243970870972, "kernel": 138.99037170410156}, "q_proj": {"bias": 2.696683406829834, "kernel": 70.16344451904297}, "v_proj": {"bias": 5.091798782348633, "kernel": 134.71148681640625}}, "feed_forward": {"intermediate_dense": {"bias": 6.082064628601074, "kernel": 234.28172302246094}, "output_dense": {"bias": 1.9151455163955688, "kernel": 215.6484832763672}}, "final_layer_norm": {"bias": 12.302684783935547, "scale": 61.214454650878906}, "layer_norm": {"bias": 2.324049949645996, "scale": 15.205697059631348}}, "23": {"attention": {"k_proj": {"bias": 6.5560832023620605, "kernel": 78.00006866455078}, "out_proj": {"bias": 2.066242218017578, "kernel": 140.90496826171875}, "q_proj": {"bias": 5.774182319641113, "kernel": 73.51222229003906}, "v_proj": {"bias": 6.0305328369140625, "kernel": 143.89315795898438}}, "feed_forward": {"intermediate_dense": {"bias": 6.21734094619751, "kernel": 201.0365753173828}, "output_dense": {"bias": 0.7316170930862427, "kernel": 125.6209945678711}}, "final_layer_norm": {"bias": 4.746160507202148, "scale": 27.713092803955078}, "layer_norm": {"bias": 1.081587553024292, "scale": 11.554892539978027}}, "3": {"attention": {"k_proj": {"bias": 0.1838577687740326, "kernel": 119.25445556640625}, "out_proj": {"bias": 1.882603645324707, "kernel": 116.36039733886719}, "q_proj": {"bias": 7.361364364624023, "kernel": 120.10975646972656}, "v_proj": {"bias": 0.8109956979751587, "kernel": 111.52893829345703}}, "feed_forward": {"intermediate_dense": {"bias": 2.0842111110687256, "kernel": 283.2494812011719}, "output_dense": {"bias": 2.0528383255004883, "kernel": 259.9748229980469}}, "final_layer_norm": {"bias": 2.24429988861084, "scale": 7.258996963500977}, "layer_norm": {"bias": 0.886063814163208, "scale": 7.4268622398376465}}, "4": {"attention": {"k_proj": {"bias": 0.7116657495498657, "kernel": 124.03801727294922}, "out_proj": {"bias": 1.6178038120269775, "kernel": 121.1064453125}, "q_proj": {"bias": 8.469348907470703, "kernel": 122.57626342773438}, "v_proj": {"bias": 1.193326473236084, "kernel": 118.79801940917969}}, "feed_forward": {"intermediate_dense": {"bias": 1.9475750923156738, "kernel": 284.6278991699219}, "output_dense": {"bias": 1.6244221925735474, "kernel": 257.72808837890625}}, "final_layer_norm": {"bias": 2.101145029067993, "scale": 7.198958396911621}, "layer_norm": {"bias": 1.1949561834335327, "scale": 6.845163822174072}}, "5": {"attention": {"k_proj": {"bias": 0.69754958152771, "kernel": 120.28215026855469}, "out_proj": {"bias": 1.8656153678894043, "kernel": 129.35032653808594}, "q_proj": {"bias": 8.441923141479492, "kernel": 119.09683227539062}, "v_proj": {"bias": 1.2337464094161987, "kernel": 129.142333984375}}, "feed_forward": {"intermediate_dense": {"bias": 2.1793618202209473, "kernel": 280.69134521484375}, "output_dense": {"bias": 1.8615474700927734, "kernel": 257.15997314453125}}, "final_layer_norm": {"bias": 2.1591551303863525, "scale": 7.1681413650512695}, "layer_norm": {"bias": 1.2301253080368042, "scale": 6.431128025054932}}, "6": {"attention": {"k_proj": {"bias": 0.267674058675766, "kernel": 123.57808685302734}, "out_proj": {"bias": 1.6781013011932373, "kernel": 120.11839294433594}, "q_proj": {"bias": 7.179095268249512, "kernel": 123.54954528808594}, "v_proj": {"bias": 1.1907923221588135, "kernel": 119.64810180664062}}, "feed_forward": {"intermediate_dense": {"bias": 3.1332929134368896, "kernel": 283.4316101074219}, "output_dense": {"bias": 1.9345507621765137, "kernel": 260.4262390136719}}, "final_layer_norm": {"bias": 2.715006113052368, "scale": 8.391536712646484}, "layer_norm": {"bias": 0.9009594321250916, "scale": 7.117555141448975}}, "7": {"attention": {"k_proj": {"bias": 0.45968830585479736, "kernel": 117.6408462524414}, "out_proj": {"bias": 2.534296989440918, "kernel": 128.3868408203125}, "q_proj": {"bias": 8.063133239746094, "kernel": 117.92467498779297}, "v_proj": {"bias": 1.522931456565857, "kernel": 126.03628540039062}}, "feed_forward": {"intermediate_dense": {"bias": 3.0588486194610596, "kernel": 278.9996032714844}, "output_dense": {"bias": 2.2587409019470215, "kernel": 270.8092041015625}}, "final_layer_norm": {"bias": 2.423645257949829, "scale": 7.401341438293457}, "layer_norm": {"bias": 1.041681170463562, "scale": 6.691577911376953}}, "8": {"attention": {"k_proj": {"bias": 0.7018513083457947, "kernel": 121.97882080078125}, "out_proj": {"bias": 3.489515781402588, "kernel": 135.91600036621094}, "q_proj": {"bias": 9.189650535583496, "kernel": 122.31185913085938}, "v_proj": {"bias": 1.8782660961151123, "kernel": 134.99163818359375}}, "feed_forward": {"intermediate_dense": {"bias": 2.7773525714874268, "kernel": 300.2033386230469}, "output_dense": {"bias": 3.839596748352051, "kernel": 287.6004638671875}}, "final_layer_norm": {"bias": 6.943517684936523, "scale": 20.858488082885742}, "layer_norm": {"bias": 0.9867479205131531, "scale": 6.735840797424316}}, "9": {"attention": {"k_proj": {"bias": 0.5994160175323486, "kernel": 121.50614166259766}, "out_proj": {"bias": 4.258487224578857, "kernel": 140.7706298828125}, "q_proj": {"bias": 9.54395580291748, "kernel": 122.70510864257812}, "v_proj": {"bias": 1.9621758460998535, "kernel": 136.7404022216797}}, "feed_forward": {"intermediate_dense": {"bias": 6.2329535484313965, "kernel": 307.13275146484375}, "output_dense": {"bias": 6.060464859008789, "kernel": 305.422119140625}}, "final_layer_norm": {"bias": 10.53576374053955, "scale": 39.37849044799805}, "layer_norm": {"bias": 0.9422033429145813, "scale": 6.8985490798950195}}}, "pos_conv_embed": {"conv": {"bias": 15.410429954528809, "weight_g": 21.016075134277344, "weight_v": 215.2329864501953}}}, "feature_extractor": {"conv_layers": {"0": {"conv": {"bias": 0.5982058644294739, "kernel": 8.08896541595459}, "layer_norm": {"bias": 10.069783210754395, "scale": 10.451257705688477}}, "1": {"conv": {"bias": 4.74075174331665, "kernel": 90.8435287475586}, "layer_norm": {"bias": 6.922820091247559, "scale": 19.5467586517334}}, "2": {"conv": {"bias": 6.7732415199279785, "kernel": 146.13897705078125}, "layer_norm": {"bias": 9.044225692749023, "scale": 19.424888610839844}}, "3": {"conv": {"bias": 5.224758148193359, "kernel": 159.10508728027344}, "layer_norm": {"bias": 8.319666862487793, "scale": 17.64743423461914}}, "4": {"conv": {"bias": 4.434978008270264, "kernel": 157.35813903808594}, "layer_norm": {"bias": 9.193974494934082, "scale": 15.562357902526855}}, "5": {"conv": {"bias": 5.297643661499023, "kernel": 131.1835174560547}, "layer_norm": {"bias": 10.735219955444336, "scale": 13.812533378601074}}, "6": {"conv": {"bias": 5.615579128265381, "kernel": 136.41822814941406}, "layer_norm": {"bias": 12.515308380126953, "scale": 11.152680397033691}}}}, "feature_projection": {"layer_norm": {"bias": 10.162856101989746, "scale": 27.727237701416016}, "projection": {"bias": 4.438727855682373, "kernel": 88.27916717529297}}, "masked_spec_embed": 26.247730255126953}}, "train/learning_rate": 6.500000017695129e-05, "train/loss": 4.594516754150391, "train/param_norm": 2534.097412109375, "_timestamp": 1649194482, "_runtime": 1859, "_step": 333, "eval/loss": 4.729986667633057, "eval/wer": 1.0118892869598302, "predictions/epoch_1": {"_type": "table-file", "path": "media/table/predictions/epoch_1_166_cd7954b6d932424bb480.table.json", "sha256": "cd7954b6d932424bb480eca029caeb05868c5b91820a6d26dbf7a4c54dea4dc8", "size": 10457, "artifact_path": "wandb-client-artifact://12mubouboytts7pbt7wnu6ny9yhwfcvdvp8d78zw30974syhfj1jcekrgrnan9c5z8iuvp45pigrbyyh6gw0vl3j9lt1rwpncgyadzuehoeqg6sn3b9xmc078cmgg1kq:latest/predictions/epoch_1.table.json", "_latest_artifact_path": "wandb-client-artifact://12mubouboytts7pbt7wnu6ny9yhwfcvdvp8d78zw30974syhfj1jcekrgrnan9c5z8iuvp45pigrbyyh6gw0vl3j9lt1rwpncgyadzuehoeqg6sn3b9xmc078cmgg1kq:latest/predictions/epoch_1.table.json", "ncols": 2, "nrows": 50}, "predictions/epoch_2": {"_type": "table-file", "path": "media/table/predictions/epoch_2_333_1efa89735ecf2ed7ea1c.table.json", "sha256": "1efa89735ecf2ed7ea1cb7855ecd16d39b9ce2279e169b10971ecc40ae1a99e5", "size": 10500, "artifact_path": "wandb-client-artifact://hmv13aqsj7ca1b0bym8484lrslb1fzajycsle78zljxltkxms7gu1idr5cl1h718qjketkopzi1ced3pmmfi4i7kqwlidw0z5g7o20oqcm2g9w3rnbhzsyzaywk5jv02:latest/predictions/epoch_2.table.json", "_latest_artifact_path": "wandb-client-artifact://hmv13aqsj7ca1b0bym8484lrslb1fzajycsle78zljxltkxms7gu1idr5cl1h718qjketkopzi1ced3pmmfi4i7kqwlidw0z5g7o20oqcm2g9w3rnbhzsyzaywk5jv02:latest/predictions/epoch_2.table.json", "ncols": 2, "nrows": 50}}
\ No newline at end of file